Multi-stakeholder Collaborations & Biomarker
Development and Implementation
By
Tingyu Liu
B.A. Neuroscience
Pomona College, 2010
SUBMITTED TO THE ENGINEERING SYSTEMS DIVISION IN PARTIAL FULFILLMENT OF THE REQUIREMENTS
FOR THE DEGREE OF
MASTER OF SCIENCE IN TECHNOLOGY AND POLICY
AT THE
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
MASSACHAUSETTS INSTITUTEI
OF TECHNOLOGY
MAY 2 '1214
JUNE 2014
LIBRARIES
cc 2014 Massachusetts Institute of Technology. All rights reserved
Signature redacted
Signature of Au thor:
_
Certified by:
00
Signature redacted
Engineering Systems Division
May 9, 2014
I
Kenneth A. Oye
Associate Professor of Political Science and Engineering Systems Division
Signature redacted
Certified by:
Certified b
7V
Tnesis Supervisor
Gigi Hirsch
Executive Director of the MIT Center for Biomedical Innovation
Thesis Supervisor
Signature redacted
Magdalini Papadaki
Research Scientist of the MIT Center for Biomedical Innovation
Thesis Supervisor
Accepted by:
Signature redacted
Dava Newman
Professr of Aeronautics and Astronautics and Engineering Systems Division
Director, Technology and Policy Program
Tingyu Liu12
Multi-stakeholder collaborations & biomarker
development and implementation
By
Tingyu Liu
Submitted to the Engineering Systems Division on May 9, 2014 in Partial Fulfillment of the Requirements
for the Degree of Master of Science in Technology and Policy
Abstract
The rise of genomic technologies and increasing interest in personalized medicine have triggered a
renewed focus on the role of biomarkers in drug development and clinical care. While many
stakeholders, including industry, regulatory agencies, patients, academia, and payers, are involved in
biomarker development and use, these stakeholders reside in functional siloes that lead to fragmented
outputs and efforts. Thus, although biomarker discovery is extensive, biomarker adoption in the R&D
process and clinical care is slow, complex, and opaque. The gap between discovery and practice needs
to be bridged by clear and consistent evidentiary standards, regulations and policies. Biomarker
adoption requires a systems effort that includes active dialogue and understanding among all
stakeholders. The level of discussions and resources (data, samples, expertise, financial, etc.) needed
also makes this field particularly fertile for collaboration.
This thesis suggests that multi-stakeholder collaborations (MSCs) can address the challenges of
biomarker adoption. It describes what has been achieved in existing prominent MSCs, extracts general
lessons on their contributions, and provides recommendations on the use of MSCs in drug development.
Specifically, MSCs have demonstrated impact across 4 key areas: 1) data generation and establishment
of standards; 2) development of scientific processes for biomarker evaluation and use; 3) the structuring
of innovative operational and organizational scaffolds for collaboration; and 4) delivery of vetted
regulatory recommendations and robust biomarkers for use by drug developers and clinicians.
Understanding the roles, evolution, and challenges of MSCs will be important in enhancing the capacity
to collaborate to address the complex challenges for biomedical innovation and healthcare outcomes.
Thesis Supervisor: Dr. Kenneth A. Oye
Title: Associate Professor of Political Science and Engineering Systems Division
Thesis Supervisor: Gigi Hirsch
Title: Executive Director of the MIT Center for Biomedical Innovation
Thesis Supervisor: Magdalini Papadaki
Title: Research Scientist of the of the MIT Center for Biomedical Innovation
Tingyu Liu13
Table of Contents
A b stract .......................................................................
. ..............................................................................
2
Acknow ledgem ents .......................................................................................................................................
5
Executive Su m m ary .......................................................................................................................................
6
Section 1: Introducing biomarkers as a key tool in drug discovery/development & patient care ...........
9
Section 2: Biomarkers in drug development and patient care: background and critical challenges..........14
Section 2.1: Term inology/No m enclature ............................................................................................
14
Section 2.2: Evidentiary Standards..........................................................................................................15
Section 2.2.1: Analytical Validity...............................................16
Section 2.2.2: Clinical V alidity .........................................................................................................
16
Section 2.2.3: Clinical U tilit y.............................................................................................................
17
Section 2.3: Regulations and policies..................................................................................................
17
Section 2.3.1: The challenge of Companion Diagnostics (CDx) .......................................................
19
2.4: Summary: The Importance of MSCs in Biomarker Adoption .......................................................
20
Section 3: The impact of MSCs in promoting biomarker adoption in drug R&D ...................................
22
Section 3.1: Prequel: Voluntary eXploratory Data Submissions (VXDS) at the regulatory agencies as an
example of collaborations across stakeholders at the global scale ...................................................
23
Section 3.2: The Biomarker Qualification Program.................................................................................25
Section 3.3: Analyzing key BQ P outputs..............................................................................................
27
Section 3.4: Conclusion and next steps ...................................................................................................
29
Section 4: MSCs' role in promoting biomarker adoption in later-stage R&D and patient care..............30
Section 4.1: The Biom arkers Consortium ............................................................................................
30
Sectio n 4 .1.1: 1-SPY 2 T rials .................................................................................................................
31
Section 4.1.2: Analyzing the outputs of the Biomarkers Consortium ............................................
33
Section 4.2: The Early Detection Research Network (EDRN) ..............................................................
34
Section 4.2.1: Analyzing the EDRN outputs .........................................................................................
36
Section 5: Lessons and recommendations to enhance the role of multi-stakeholder collaborations
(MSCs) in biomarker development and implementation ......................................................................
38
Section 5.1: Creating data and data standards.......................................................................................38
Section 5.2: Developing new scientific processes..............................................................................
38
Section 5.3: Operationalizing and organizing innovative collaborations............................................39
Section 5.4: Delivering tangible outputs ...........................................................................................
40
Tingyu Liu14
Section 5.5: Recommendation-Avoid fragmentation and duplication of pre-existing efforts......40
Section 5.6: Recommendation- Maintain flexible scientific processes ............................................
41
Section 5.7: Recommendation-Strategic and timely inclusion of more stakeholders..................... 42
Section 5.8: Recommendation-Develop metrics to track progress and measure impact................43
Section 5.9: Conclusion ...........................................................................................................................
44
Glossary A: The main biomarker categorizations proposed by field experts ........................................
45
F D A ..........................................................................................................................................................
45
In d u stry ...................................................................................................................................................
46
Thomson Reuters ....................................................................................................................................
46
The Biomarkers Definitions W orking Group ......................................................................................
47
Glossary B: Mapping differences in semantics on the terms validation and qualification.................... 48
Appendix A: Background information on the MSCs that have participated in the BQP......................... 50
Appendix B: Expert biographies..................................................................................................................52
B ib liog ra p hy .................................................................................................................................................
55
Tingyu LiuI5
Acknowledgements
First and foremost, I would like to express my sincere gratitude to my advisors, Dr. Ken Oye, Dr. Gigi
Hirsch, and Dr. Magda Papadaki, for their support and guidance throughout my research and my
Master's studies. Special thanks goes to Dr. Papadaki, who patiently edited and re-edited my writing and
always provided insightful comments. I would like to thank the members of the Center for Biomedical
Innovation; my research would not have been possible without their uplifting words and banter.
I am also grateful for the subject matter experts who have kindly provided inputs and advice during the
research and writing process.
I would like to thank my fellow Technology and Policy students for the community throughout my two
years at MIT. I would especially like to thank the students who spent countless hours with me at libraries
or cafes during the last few weeks before the thesis deadline. Finally, I would like to thank my loved
ones, who have supported me fully throughout this entire process.
Tingyu Liu16
Executive Summary
Recent advances in genomic technologies have given rise to interest in new therapies that can develop
safer, better drugs for patients in an affordable and efficient manner. In particular, the biomedical
industry is devoting increased attention to the promises and potential of personalized medicine.
Personalized medicine is the effort to prevent, diagnose, and treat disease with targeted therapies
designed to be effective for individual patients based on genetic or other information.
The application of well-defined biomarkers is the cornerstone of personalized medicine. A
biomarker is an indicator of biological processes or pharmacologic responses to drug therapy.1
Identifying the presence, absence, or amount of a biomarker in a patient can predict disease
susceptibility and guide tailored therapies by assessing individual benefit or risk towards a particular
drug, or determining the optimal drug dose base on metabolism. Although genomic technologies have
brought about a rapid increase of novel biomarkers, developing accepted terminologies and scientific
standards, as well as thoroughly evaluating the biomarkers, is a slow, demanding and complex process
that comes with a variety of challenges. The novelty that accompanies many genomics biomarkers
presents a crucial regulatory challenge. Regulations around biomarkers and other cutting edge R&D
innovations, are unclear and controversial, which both hinders their adoption and may lead to
biomarker misuse. Reimbursement and pricing are also risky, since payers are still unclear and
inconsistent on evidence requirements for coverage decisions.
Most importantly, biomarker adoption in pharmaceutical development and clinical practice is a
systems effort that necessitates active dialogue and understanding among all the stakeholders, including
industry, regulatory agencies, patients, academia, and payers, as well as consolidated outputs and
efforts across the spectrum. The level of resources (data, samples, expertise, financial, etc.) needed also
makes this field particularly fertile for collaboration. In this thesis, we argue that multi-stakeholder
collaborations (MSCs) are useful tools to address the key challenges of biomarker adoption. Starting
from the scientific challenges addressed by key MSCs, we expand to describe their approaches and
achievements on novel processes and policies toward an increased systems capacity for therapeutic
innovation through biomarker adoption.
We discuss the evolution of the first regulatory pathway, termed the Biomarker Qualification
Process (BQP), to promote biomarker adoption in R&D. The development of the BQP was initiated by
one-on-one discussions between industry and the U.S. FDA (Food and Drug Administration) on how to
best integrate pharmacogenomics data into drug submissions. These discussions led to the formation of
key MSCs that mediated data-sharing and analysis agreements and resulted in the qualification of
several biomarkers for preclinical safety testing and clinical trial selection in Alzheimer's disease by both
the FDA and the EMA (European Medicines Authority).
Thus far, MSCs involved in the BQP have mainly focused on early-stage R&D biomarkers. Other
MSCs are focusing on biomarkers involved in later-stage drug R&D and patient care, including the
development of in vitro diagnostics (IVDs). Significant achievements of the Biomarkers Consortium and
the Early Detection Research Network (EDRN) include, respectively, a novel adaptive biomarker-driven
screening trial for therapeutics and a rigorous process for evaluating biomarkers for early detection.
To enhance the understanding of collaborations in biomarker adoption, we developed a
framework to distill key lessons and analyze the contributions of these key MSCs across four areas that
involve important translational enablers for biomarker adoption, particularly data and standards,
operational elements around novel scientific processes, key organizational aspects of collaborative
Note that biomarkers can be used as a tool to help drug developers during the preclinical and clinical stages of
R&D. A
biomarker can also be used as a product, in which case it is a medical device termed an In Vitro Diagnostic (IVD). An IVD can be
independent of a drug or it can be drug-specific.
IVDs that must be paired with a specific drug because detecting and measuring
this biomarker is essential for the clinical use of the drug is more precisely termed a companion diagnostic (CDx).
Tingyu Liu17
innovation, as well as the prominent tangible outputs around novel biomarkers and relevant regulatory
guidelines. In specific:
1)
Data and data standards. Data sharing allows MSCs to tackle projects that would not
otherwise be feasible. Moreover, broad-based data harmonization can set important
standards to increase consistency in cross-stakeholder interpretations, as well as subsequent
stages of data deployment and decision-making such as regulatory submissions, the design of
novel clinical trials, or drug labelling and use.
2) Scientific processes. The scientific processes developed by MSCs are essential to break down
large, complex challenges into manageable workstreams. MSCs have been pivotal in the
establishment of new pathways for biomarker development and use across R&D stages and
disease areas. For example, the MSCs involved in the BQP are collaborating to move qualified
preclinical biomarkers into the clinical stage. In addition, the I-SPY 2 trial for early-stage
breast cancer has led to new adaptive trials for lung cancer and Alzheimer's disease. Such
follow-on work will increase the predictive capacity and therapeutic impact of biomarkers.
3) Collaborative frameworks. Biomarker development is a complex field that requires an
increasing degree of knowledge communication and integration across stakeholder groups
and R&D stages. Learnings from the MSCs analyzed in this thesis depict important
developments in the operational and structural scaffolds of collaborative innovation,
allowing the design and execution of the scientific processes discussed above and their
ultimate deliverables. MSCs have continued to evolve in scope and breadth through strategic
partnerships and interactions; their follow-on work will create progressive new value
generation from biomarker and related translational outputs. Most notably, MSCs have
increased the level of understanding and alignment on important strategic directions and
priorities, within and across the different stakeholder groups.
4) Concrete deliverables. A number of robust biomarkers and regulatory guidance documents
that outline innovative policies for biomarker adoption have already emerged from the work
of MSCs. A few of the prominent examples of progress include nine qualified nephrotoxicity
biomarkers, two qualified biomarkers for Alzheimer's disease, and five FDA-approved
diagnostics for early-stage cancer detection. Guidance documents include a regulatory
pathway for qualifying biomarkers (the BQP) and methods to perform adaptive clinical trials.
MSCs have already contributed to innovative approaches to tackle the challenges of biomarker
adoption in drug R&D and clinical care. Our distilled learnings about their contributions have led to a
number of key recommendations that will be important as MSCs continue to accelerate biomarker
innovation. The recommendations can be broadly summarized as:
1) Avoid fragmentation and duplication of pre-existing efforts. MSCs have broken down
complex biomarker research challenges and brought together different stakeholders. As
MSCs continue to evolve, it is imperative to pursue a higher level of collaboration by
bringing together different MSCs to share key knowledge and processes. Such an effort will
insure that MSCs are aware of and can leverage each other's contributions across the four
areas described in the above framework.
2) Keep scientific processes flexible and adaptive. MSCs have already established key
scientific processes for biomarker development and use. It is essential to keep the processes
adaptive, with the ability to expand their scope as more data and information becomes
available. An important enabler in this process will be regulatory oversight on the ground of
flexible guidance documents that can readily incorporate new learnings in the field.
3) Progressively include stakeholders. MSCs in the biomarker field have mainly included
regulators, industry, and academia thus far. However, our study showed that other
stakeholder groups, including diagnostics companies, payers, and patients, need to be more
Tingyu Liu18
systematically included in MSCs because successful biomarker development and use
requires their contributions and insights as well. Depending on the goals of an MSC,
regulators and industry may join at different times. For example, an MSC that focuses on
scientific research as opposed to commercial outputs may choose to engage industry later in
the biomarker development process.
4) Develop metrics to track progress and measure impact. There is currently a lack of
meaningful metrics and performance measurement approaches for collaborative efforts. A
key barrier that is hindering metrics development in the biomarker field is the
confidentiality provisions as biomarkers move back into the commercial space. MSCs should
discuss ways to remove such barriers since meaningful evaluation is essential to set goals,
identify gaps, and optimize collaborations.
The biomedical industry has gradually shifted towards a more collaborative approach to tackle
challenges that exceed individual stakeholder capabilities and integrate the current fragmented
paradigm in drug R&D. As science and technology become more diffuse and specialized, MSCs will be
integral for successful therapeutic innovations. This thesis has charted the progress and contributions of
key collaborations in the biomarker field to elucidate key learnings and future recommendations. Such
analysis will help MSCs evolve into established operational models for innovation in biomedicine and
healthcare.
Tingyu Liu|9
Section 1: Introducing biomarkers as a key tool in drug
discovery/development & patient care
Recent advances in biotechnology, prompted by the sequencing of the human genome, have resulted in
a new array of therapeutics. Indeed, the rise of the so-called Omics Revolution (genomics,
proteomics,
pharmacogenomics) and the -ics technologies (informatics, bioinformatics, biostatistics) have expanded
the ways in which disease is detected and treated. There are now over 30,000 primary genes and over
100,000 proteins and many of their protein and metabolic products serve as promising new molecular
targets (Day et al. 2009).
To turn these possible targets into real treatments requires a complex, systemic and highly
regulated process. In the United States, the main authority that regulates drugs is the Food and Drug
Administration (FDA), and especially the Center for Drug Evaluation and Research (CDER). In the
European Union, the FDA counterpart is the European Medicines Agency (EMA). Pharmaceutical R&D
has traditionally followed four main stages (Figure 1):
Figure 1. Stages of pharmaceutical R&D.
On average, this process takes more than ten years, costs more than a billion dollars, and only one in ten
thousand compounds actually succeed to become a marketed drug (Paul et al. 2010). Many experts
have studied R&D productivity and success to propose solutions on improving the process (Pammolli et
al. 2011).
The capacity of the healthcare system as a whole to provide newer, better drugs to patients in
an affordable, reliable, sustainable, and efficient manner has been hindered by a series of bottlenecks in
drug R&D and patient care (Table 1). The challenges are already apparent in the discovery stage; the
proliferation of new molecular targets due to the genomics revolution is a key example. These targets
are frequently so novel that there is insufficient data and/or a lack of harmonized data standards to
understand and further assess the target in question. Another bottleneck is the high attrition rate of
late-stage clinical trials (Carroll 2012); there is a lack of robust safety or efficacy profiling because of an
imperfect understanding of the disease and the science, including the drug target and drug design. This
has been particularly evident in the field of oncology; drug developers have struggled to obtain
significant clinical results when the heterogeneity of the patients' genotypes and clinical presentation
increase in later-stage clinical trials (Gerard et al. 2012). On the other hand, a drug may fail to reach a
patient because of reimbursement requirements by payers which have become increasingly uncertain
and stringent, especially in the light of the Affordable Care Act (Trusheim & Berndt 2012) and recent
global fiscal pressures. A new drug may have very limited value for patients if insurance companies
Tingyu Liu 110
refuse to cover the drug because they believe that it is not sufficiently superior to existing therapies or
its cost-benefit analysis is not sufficiently robust.
Table 1. The healthcare value chain for drugs and devices has been hindered by a series of
bottlenecks.
"
"
"
Lack of clear criteria to prioritize targets
Lack of harmonized data standards
Lack of good animal models for diseases
"
Unclear safety and toxicity profile of drugs
"
Inability to demonstrate safety and efficacy
*
Inability to scale-up drugs/devices for large-scale manufacturing
"
"
"
Different decision criteria among stakeholder groups
Heterogeneity of patients leads to clinical trial failures
Safety and efficacy can only be optimized for patient sub-populations
*
"
Limited patient access due to high prices and/or lack of reimbursement
Unknown cost-effectiveness evidence compared to standard of care or for novel
"
"
treatments
Long timelines
High development costs
"
Low probability of success
"
Lack of communication & fragmented actions between the different
stakeholder groups
The formal definition of biomarkers commonly used today was established by the Biomarkers
Definitions Working Group's workshop in 2001. By this definition, a biomarker is "a characteristic that is
objectively measured and evaluated as an indicator of normal biological processes, pathogenic
processes, or pharmacologic response to a therapeutic intervention" (Atkinson et al. 2001). Moreover, a
"characteristic" is any "biological molecule found in blood, other body fluids, or tissues that is a sign of
normal or abnormal process, or of a condition or disease and may be used to see how well the body
responds to a treatment for a disease or condition." The potential value of the application of biomarkers
in clinical care has been known for thousands of years. For example, in 4th to 5th century BCE, ancient
clinicians detected glucose in urine by observing that urine from certain individuals attracted ants, thus
providing a clinical description of diabetes. However, in the past twenty years, the term "biomarker" has
come into focus and popularity, especially after the sequencing of the human genome (Vaidya &
Bonventre 2010). Biomarkers are a key technology to address the bottlenecks which are preventing
patients from obtaining innovative therapies and especially personalized medicines.
Biomarkers serve as the foundation of personalized medicine, an emerging healthcare approach
that recognizes the importance of individual differences in treating disease. It is beginning to be
understood that there are multiple subgroups and stages for each disease, and traditional drug
development is not fully considering the development of targeted drugs tailored to a specific population
subgroup. Personalized medicine aims to prevent, diagnose, and treat disease based on a thorough
understanding of the genetic basis of the disease and its clinical presentation in individual patients.
Identifying the presence, absence, or amount of a biomarker in a patient can predict disease
Tingyu Liu 11
susceptibility and guide tailored therapies by assessing individual benefit or risk towards a particular
drug, or determining the optimal drug dose base on metabolism.
A classic example in the field of personalized medicine is trastuzumab (Herceptin, developed
by
Genentech). Herceptin is a molecularly targeted cancer therapy that is designed to "shut off" the
HER2
gene. The HER2 gene is responsible for the overexpression of the HER2 protein in approximately
20-25%
of breast cancer cases and results in a very aggressive form of the disease. Thus, a breast cancer
patient
is first tested for HER2 using an in vitro diagnostic (IVD). The patient can only use Herceptin after
obtaining a positive test result because the drug is only beneficial for patients with this genotype
(FDA
2013b).
In this thesis, we have defined biomarkers to assume three main roles (Table 2).2 Biomarkers
used in R&D are typically basic assays and tests to help drug developers understand their drug
and its
effects in both the preclinical and clinical stages. However, sometimes a biomarker used in R&D
can be
further developed as a companion diagnostic (CDx), a subset of IVDs. This is a biomarker that is
matched
to the drug because detecting and measuring this biomarker is essential for the clinical use of
the drug,
as the Herceptin example illustrates. Note that a CDx does not have to be developed during drug
R&D.
IVDs which are not CDx are used independently of a drug (see examples below, Table 2).
Table 2. Examples of the different uses of biomarkers.
1) Identify targets for therapy, e.g. circulating mRNA as signaling
molecules
2) Demonstrate drug safety, e.g. by measuring upregulation of KIM-1
biomarker to predict toxic kidney injury in animal models
1) Determine breast-cancer patients' therapy based on HER2 profile for
Herceptin therapy
2) Determine melanoma patients' therapy based on presence of V600E or
V600K mutation in BRAF gene for Tafinlar and Mekinist
1) Oncotype DX - 21-gene panel test that predicts likelihood of cancer
recurrence in women who are estrogen receptor (ER) positive
2) CertNDx - diagnostic test for bladder and upper urinary tract cancers by
detecting FGFR3 mutation in urine
3) 23andMe - direct-to-consumer (DTC) test that assesses inherited traits,
genealogy and possible congenital risk factors
The value that biomarkers bring to the R&D and, correspondingly, to patient care is evident
(Table 3). Biomarkers have the potential to overcome the bottleneck examples in Table 1 above
and
thus significantly reduce the costs and uncertainty associated with drug development. Biomarkers
are
useful from the earliest stages of drug development, to identify drug targets and establish how
the
potential drug interacts. Biomarkers help elucidate how the drug behaves in animal models and human
patients, and eventually define the target population. As mentioned previously, if a drug
was elucidated
2 Note
that while these roles are commonly understood in the biomarker field,
biomarkers are not typically
thought of and categorized in such a manner because experts often only discuss one aspect of
a biomarker's use,
as opposed to all three.
Tingyu Liu |12
with the use of a biomarker, then the biomarker can become a CDx. Moreover, biomarkers are useful to
detect disease and understand a patient's profile as an lVD.
Table 3. Value that biomarkers bring to R&D and patient care.
Biomarkers deliver value to each stakeholder across drug R&D stages and clinical practice to
help experts make more informed decisions that will better apprise a patient's clinical profile and/or
deliver safer, more effective drugs to patients (Table 4). Pharmaceutical companies use biomarkers to
help demonstrate the safety and efficacy of new drugs. Some, such as Roche and Novartis, are building
dedicated in-house diagnostics units while others such as AstraZeneca are forming partnerships with
external developers (Opar 2011). Diagnostic companies can partner with pharmaceutical companies or
launch diagnostic devices by themselves. Thus, it is of no surprise that the global diagnostics market is
$42 billion (with $16 billion in the US in 2011) and over 6.8 billion diagnostic tests are performed each
year (Advameddx 2013). Regulatory agencies have also prioritized biomarker research and development
to better understand how to evaluate biomarker use and to promote biomedical productivity
in a
(Woodcock & Woosley 2008). Payers and providers want to understand if patients are being treated
way that takes into account their individual differences, to maximize efficacy of cures and save both
money and lives.
Tingyu Liu 113
Table 4. Biomarkers deliver value to each stakeholder in the healthcare value chain.
Still, biomarkers uptake comes with many challenges. This thesis discusses the challenges as well
as the opportunities of biomarker integration into drug development and patient care. We then argue
that multi-stakeholder collaborations (MSCs) can be seen as a useful instrument to address the complex
and costly challenges that are currently constraining productive use of biomarkers. Successful biomarker
adoption requires a systems-level change that involves and impacts all stakeholders (Barratt et al. 2012;
Stephenson et al. 2013; Chataway et al. 2012; Goldman 2012; Wagner et al. 2007; Campion et al. 2013).
The term "MSCs" (as distinct from "consortia" or "collaborations" in general) is defined as an
environment that convenes two or more different stakeholder groups, such as the ones listed in Table 4.
Abstracting from our research in the biomarker field, we make a case for the role of MSCs beyond
scientific outcomes in the design and implementation of effective policies which will allow all
stakeholders to incorporate emerging technologies and science innovation in drug discovery and
healthcare in more sustainable and strategic ways.
Tingyu Liu 114
Section 2: Biomarkers in drug development and patient care:
background and critical challenges
The challenges to bring biomarkers into R&D and patient care span from nomenclature to policy; the
details are complex and it is difficult for invested stakeholders, even field experts, to understand the
whole picture and make informed decisions. In this section, the challenges are broken down into three
aspects: the confusion around terminology/nomenclature, the complexity of evidentiary standards, and
controversies in regulations and policies. It is important to note that each of these three aspects feeds
into the other, and they are not mutually exclusive. This section ends with a summary of these
challenges in the context of a biomarker's role in pharmaceutical R&D, as a CDx, or as a
diagnostic/prognostic device.
Section 2.1: Terminology/Nomenclature
The definition of a biomarker, as discussed in Section 13, although detailed in content, is overly simplistic
in concept, failing to illustrate the many different types and categorizations of biomarkers. The present
study contains a Glossary that lists the most common terminologies used by various stakeholders to
depict the current confusion in the biomarker landscape (Glossary A). The confusion is mainly due to the
many different groups working in this field, each with their own agenda and points of focus.
It is important to note that for a biomarker to be used for any intended purpose, it needs to be
measured with a corresponding device. The device is a way to assay or illustrate the state of the
biomarker (Woodcock 2010).4 Devices are regulated by the Center for Devices in Radiology and Health
(CDRH), as opposed to CDER. One of the fundamental complexities in the field is that regulatory
evaluation considers the biomarker separately from its measurement device, although in practice, using
a biomarker necessitates an instrument or tool to measure it. The comprehensive definition of a
biomarker with its corresponding measurement device highlights the complexity of the field and the
reason why definitions and categorizations fail to capture all biomarkers. The term "biomarker" is vast,
including any imaging technique (CT, MRI, PET, etc) and any clinical laboratory test, from
immunochemistry, blood pressure, pain scales, and microbial cultures to ECGs. This situation is further
complicated by the recent abundance of novel biomarker discoveries resulting from the Human Genome
Project (Gutman & Kessler 2006).
The confusing nature of biomarker terminology extends to complicate the evaluation framework
for assessing biomarkers. There is a common tendency to overuse or misuse the terms
validation/qualification, as well as the terms analytical validity, clinical validity, and clinical utility.
Glossary B provides examples of how different experts have defined the terms validation and
qualification. Experts from different stakeholder groups tend to use the terms loosely and may have
their preferred definition of the word since each group has its own areas of focus and decision-making,
As discussed in Section 1, the formal definition of biomarkers as established by the Biomarkers Definitions
Working Group is "a characteristic that is objectively measured and evaluated as an indicator of normal biological
processes, pathogenic processes, or pharmacologic response to a therapeutic intervention" (Atkinson et al. 2001).
Moreover, a "characteristic" is any "biological molecule found in blood, other body fluids, or tissues that is a sign
of normal or abnormal process, or of a condition or disease and may be used to see how well the body responds to
a treatment for a disease or condition."
4 A device is formally defined as "an instrument, apparatus, implement, machine, contrivance, implant, in vitro
reagent, or other similar or related article, including any component, part, or accessory, which is...intended for use
in the diagnosis of disease or other conditions, or in the cure, mitigation, treatment, or prevention of disease, in
man or other animals...and which does not achieve its primary intended purposes through chemical action within
or on the body of man or other animals and which is not dependent upon being metabolized for the achievement
of its primary intended purposes" Section 201(h) of the Federal Food, Drug, and Cosmetic Act, 21 U.S.C 321(h).
3
Tingyu Liu 115
languages, and norms, as well as different valuation of the data (EGAPP 2014). Other terms that have
evolved into use seem to be derivations of these terms, including analytical validation or analytical
performance, clinical qualification, utilization or clinical utilization.
Despite the overall confusion, seminal papers by industry authors and the FDA are in general
agreement and have been widely cited and these terms will be used for the purposes of this thesis.' On
this ground, after the initial biomarker discovery, there are three general stages of biomarker
development:
1) analytical validity-validated method/assay performance;
2) clinical validity-validated link between the biomarker and biological/clinical endpoints;
3) clinical utility-the relevance and usefulness of the biomarker in patient care.
Section 2.2: Evidentiary Standards
Biomarker evaluation and acceptance is very slow compared to the speed of biomarker discovery
(Bleavins 2010). There has been no shortage of novel biomarkers (Rinaldi 2011), but it is difficult to
determine which ones have the greatest potential to be further developed for use in drug development
and patient care. This difficulty arises from the fact that the key stakeholder groups work in functional
silos of data and knowledge generation. Research labs, where most biomarker discovery and early
development occurs, only have limited communication with and awareness of other labs. Moreover,
industry and diagnostic companies are not aware of relevant research pursued at academic labs.
Conferences, publications, and university technology transfer offices remain the key means of
communication between stakeholders, although this is a slow and unreliable route. There is no
systematic way of linking the stakeholders and promoting cross-stakeholder awareness of each other's
work and needs. Thus, although academics may lead on important biomarker discoveries, they often do
not understand the rigor needed to move a biomarker through the different stages of maturity.
To ensure that the biomarker has practical use beyond its initial discovery, the gap from
discovery to practice needs to be bridged by consistent and well-accepted evidentiary standards that
can be systematically applied to every novel biomarker within reason. Evidentiary standards are a set of
parameters that underlie a biomarker's lifecycle as it steps through progressive stages of maturity, from
analytical validity to clinical validity and clinical utility. A defined set of parameters does not yet exist,
although field experts have agreed that certain key characteristics are important to keep in mind.
First, evidentiary standards depend on the intended use of a biomarker; a biomarker used in
early stage R&D for an exploratory or observational study and/or for internal decision-making may not
require the same evidentiary standards as a biomarker that is used for regulatory decision-making
and/or for treatment decisions in clinical practice such as a CDx (Lesko 2007; Lesko & Atkinson 2001;
Bleavins 2010). As the Biomarker Definitions Working Group stated, "robust linkage of a biomarker with
a clinical endpoint is not essential in early development when the goal is confirmation of pharmacologic
activity or optimization of dose regimen" (Atkinson et al. 2001). It is important to note that evidentiary
standards are science- and context- specific, so they must be reevaluated if a biomarker is used in a
different way, or in another medical context (Altar et al. 2008). Thus, finally, it has also been generally
agreed by industry and academics alike that biomarkers should be reevaluated on a continual basis as
more data is generated. This dynamic and resource-intensive nature of biomarker evolution is difficult to
sustain. Multi-stakeholder collaborations are useful for such maintenance responsibilities because a
single stakeholder cannot tackle such a large and multidisciplinary challenge.
s For a historical review of papers that detail the evidentiary standards of biomarkers, (IOM 2010a) gives special
attention to nomenclature, systems of classification, and statistical methods developed for their evaluation.
Tingyu Liu 116
Experts have been struggling with how to define the set of parameters for each evidentiary
standard to bring biomarkers from the discovery laboratory to practical use in the healthcare value
chain. There are no set definitions and standards, and no agreed-upon regulations that explicitly
evaluate a biomarker as it progresses through subsequent stages of maturity (Khleif et al. 2010).
Determining which biomarker possesses clinical validity or utility is agreed on a case-by-case basis. This
is one of the key issues that results in the slow integration of biomarkers into mainstream patient
management (Wilson et al. 2007). Below, the issues surrounding each stage of biomarker development
are highlighted in greater detail.
Section 2.2.1: Analytical Validity
Analytical validation refers to the assessment of the biomarker's measurement device and its
performance (Bleavins 2010). It has been said that analytical validity is "often an underappreciated
component of biomarker evaluation" (Pennello 2013). Even the most basic assays, such as
immunohistochemistry and fluorescence in-situ hybridization tests, can produce varying results
depending on who is performing the assay, where the reagents and samples come from, and how the
samples were prepared and stored (Gutman & Kessler 2006). This issue becomes magnified with novel,
genomic biomarkers because technology is rapidly evolving and interpretations are not standardized or
widely comprehensible (EGAPP 2014).
Another issue is inconsistent results across laboratories; laboratories may have different
protocols for the same test or different quality systems, especially if they are not certified from the
same auditing organizations (Wilson et al. 2007; Halim 2009; Khleif et al. 2010). Sample variability is yet
another issue, as many errors and inconsistencies can occur in the pre-analytical phase and affect
sample integrity and analysis suitability. For example, patients may have interfering medications or
different nutritional diets, technicians may encounter organizational issues such as mislabeling a sample
with the wrong ID, or one sample could be processed and stored differently from another (Drucker &
Krapfenbauer 2013).
Many of these issues are due to a lack of industry agreement on validation standards and
criteria in the field (IOM 2010a; Khleif et al. 2010) and the fact that much of the research concerning
analytical validity is unpublished or proprietary, making it even more difficult to synthesize the data
(EGAPP 2014).
Section 2.2.2: Clinical Validity
Table 3 (Section 1) defines the three main roles of a biomarker as a tool used in R&D, a CDx, or a drugindependent diagnostic device. Each biomarker use has its own set of evidentiary standards, and
analytical method validation must be demonstrated for all of these roles. However, biomarkers used in
the early stages of R&D, such as during the drug discovery phase, may not need to demonstrate clinical
validity. To the contrary, biomarkers used in later stages of R&D, or as IVDs, should demonstrate clinical
validity as well as clinical utility if possible, because the biomarker will be used in patient management.
Clinical validity assesses the link of analytical results to clinical and biological endpoints. An
example of a clinically valid biomarker is the count of CD4+ immune system cells for HIV and AIDs.
Clinical validity comprises a critical point of disconnect among the different stakeholder groups. A
research lab typically performs the preliminary analytical validity testing, but they often do not have a
grasp of what is needed in terms of sample size, test cross-validation, biospecimen quality, and the
translation of the biomarker to a tangible product. Thus, their data will be of limited use downstream for
industry and regulators. Indeed, one expert has stated, "Published academic studies...typically lack the
scale, analytical stringency, and level of evidence needed to elicit industry confidence to invest in
expensive clinical validation trials" (Poste 2012). Other stakeholder groups may not have a full
understanding of the scene either. In many cases, especially with genomic technologies, the FDA does
Tingyu Liu 117
not have a complete idea of what they are looking for, and drug developers are unclear on how to
proceed (Khleif et al. 2010).
Meanwhile, clinical validity standards are being raised. Ideally, clinical validity standards would
be determined by comprehensive retrospective studies and prospective clinical trials (Teutsch et al.
2009). However, such trials require significant resources and industry is hesitant to provide funding
because it is unclear how regulators would evaluate the trials. Moreover, obtaining a comprehensive
picture of the biomarker may be difficult, impossible, or unethical if it requires recruiting biomarkernegative participants for a prospective trial. For example, a biopsy has never been performed on a PSA
(Prostate Specific Antigen)-negative individual for the purposes of completely validating the biomarker
(Gutman & Kessler 2006). Also, errors from the analytical validity data can spill over and cause serious
consequences for clinical validity studies if they remain unrealized or undetected (Baggerly & Coombes
2011). For example, in the case of a CDx, the drug development program can take ten years, making the
use of different biomarker assay platforms or the switching of reagent vendors very likely (Halim 2011).
Section 2.2.3: Clinical Utility
Clinical utility assesses the quality and quantity of the generated data from the analytical and clinical
validity studies, as well as its consistency and generalizability, to determine whether the biomarker is
robust enough to be used in patient care for a specific context of use (COU) (Teutsch et al. 2009).
However, generating sufficient, high-quality evidence is difficult due to resource constraints and
uncertain market incentives (Parkinson et al. 2014). For example, at the National Cancer Institute (NCI),
much of the financial resources are set for biomarker discovery, and less than 2% of the funding is
focused on clinical utility (Schully et al. 2011). The need for both increased market incentives and
sufficient evidence is a catch-22 problem: without sufficient evidence, there is no market incentive, and
without an incentive, no evidence will be generated. Experts have stated that evidence for clinical utility
should be identified by multiple stakeholders to enable efficient generation of the needed data (Zineh &
Lesko 2009). Patient and public advocacy groups can also push for clinical acceptance of such
biomarkers. However, without a formal patient inclusion process in place, they represent an
"idiosyncratic process that depends on the emergence of leadership in a given disease area" (Mittleman
et al. 2013).
There is still no general consensus on what clinical utility should involve. Some organizations,
such as the National Institute for Health and Clinical Excellence (NICE) in the UK, and the CDC's
Evaluation of Genomic Applications in Practice and Prevention (eGAPP) group have tried to define
clinical utility (Poste et al. 2012). Some organizations believe that healthcare economics should be part
of the evaluation process, which makes the concept of clinical utility more rigorous. Clinical utility can
also be surprisingly elusive. Some of the most widely and extensively used biomarkers, like mammogram
images and the PSA, are good case studies of this ambiguity. Although these biomarkers may be
analytically and clinically valid, there is still no clear evidence that they lead to better health outcomes
(Trusheim et al. 2013).
Section 2.3: Regulations and policies
When biomarkers become products that are used in patient care, they are termed IVDs and are
regulated by the FDA as medical devices. This feeds into the challenges of biomarker adoption on the
regulatory front, because the regulations around medical devices that measure biomarkers are very
unclear and controversial, which may lead the biomarker to be misused in patient care.
Tingyu Liu118
There are two paths that a biomarker device can take. The first path goes through FDA's
approval process for medical devices. 6 FDA-approved IVDs will demonstrate clinical validity (Kelloff &
Sigman 2012), but will demonstrate clinical utility if it is an inherent part of clinical validity, as is the case
with CDx, since its performance predicts patient outcomes (Parkinson et al. 2014). On the other hand, a
device may not have to undergo FDA approval if it is considered a Laboratory Developed Test (LDT).'
Both IVDs that are reviewed by the FDA or offered as an LDT are intended for clinical use and so need to
be performed in laboratories which are inspected, certified, and accredited under the Clinical Laboratory
Improvement Amendments (CLIA), administered through CMS (Parkinson et al. 2014).8 However, CLIA
focuses on technical performance and doesn't address clinical validity or utility (Trusheim et al. 2013).
Thus, many experts regard LDTs as a "loophole" for industry to bypass the long and expensive FDAapproval pathway (Khleif et al. 2010).
LDTs can make the claim of clinical validity and utility and, with the right marketing forces, be
used in many hospitals and doctor offices. Because many healthcare stakeholders work in functional
siloes, doctors and hospitals may not understand the difference between an LDT and an FDA-approved
test. LDTs can even mimic FDA-approved tests and erode the latter's market share (Hayes et al. 2013).
Furthermore, a new type of LDTs, Direct-to-Consumer (DTC) tests such as 23andMe, pose additional
problems, because most patients will not understand the scientific uncertainty and validity surrounding
these tests well enough to make informed decisions, resulting in the possibilities of biomarker misuse
(Kelloff & Sigman 2012).
However, LDTs are useful in a variety of cases. For example, the LDT pathway allows the
evaluation of a cutting-edge technology before embarking on the costly and time-intensive FDA
approval process (Hockett & Close 2010). In addition, sometimes patient groups may advocate for faster
access for tests instead of waiting for well-designed clinical trials to demonstrate validity and utility,
depending on the urgency of their disease (Teutsch et al. 2009).
The current proliferation of diagnostics has been frequently accompanied by high price
demands, especially with biomarkers that use cutting-edge genomic technologies. In response, payer
organizations are starting to demand demonstrates of clinical utility before reimbursing such devices.
However, payers typically do not provide guidance as to what kind of data and how much data are
6
The Medical Device Amendments to the Food, Drug, and Cosmetic Act in 1976 gave the FDA authority to regulate
medical devices. The FDA established 3 classes for medical devices:
1) Class I have the least amount of risk and typically don't require premarket review or approval before they
are used in patient care.
2) Class I have moderate risk and need to obtain 510(k) approval. 510(k) approval doesn't necessarily
require clinical trials (Evans 2010), but does require the device manufacturer to show proof that the
device is equivalent to a device already on the market.
3) Class Ill have the highest risk and have to undergo the PMA process which includes the submission of
supporting clinical trial data (Weiss 2012). Note that in this case, an investigational device exemption (IDE)
application may need to be filed. An IDE permits use of the device in a clinical investigation to evaluate
the safety and/or efficacy of the investigational medical device.
Originally, LDTs were well-characterized, low-risk diagnostics that were developed in one lab to be used only in
that lab. Thus, FDA has exercised discretion in enforcing these tests. With the molecular and genetic science
revolution, these tests became more complex, utilizing epigenetic profiling and single nucleotide polymorphism
(SNP) analysis, and are increasingly used to assess high-risk diseases to inform critical treatment decisions even if
the results do not actually support the decision (Hockett & Close 2010).These tests are also now offered by large
commercial laboratories that provide services for many institutions (CDRH 2010).
8 An exception to this rule are IVDs that are intended for over-the-counter (OTC) use by the patient and IVDs that
are used at the point of care (POC), by the attending physician outside of the laboratory (Gibbs 2010).
Tingyu Liu 119
needed to support their decision to reimburse medical devices, so industry is not sure how much testing
and what kind of testing is required (Quinn 2010). Simultaneously, most LDTs lack sufficient evidence of
their impact on healthcare outcomes, so providers and payers are unsure how to use the tests and how
much they should pay (Hayes et al. 2013). FDA approval is one way to demonstrate value, but even FDAapproved devices may not be reimbursed. FDA has tried to help resolve this issue by introducing a pilot
process where a diagnostic can pursue regulatory and reimbursement approvals at the same time, but it
has not been widely useful or used (Hollmer 2013).
Section 2.3.1: The challenge of Companion Diagnostics (CDx)
An important subsection of biomarkers are the CDx, which have substantial challenges on their own. As
discussed in Section 1, a CDx is a biomarker product that is essential for the safe and efficacious use of a
drug. A CDx can be co-developed with a drug (the ideal case), or to "rescue" a drug that demonstrated
spotty performance during trials, or it can be developed after the drug is on the market (in which case
the drug label will be updated accordingly). CDx are high-risk products because the failure of a CDx is
equal to the risk of drug misuse.
In most cases, the
pace of drug development
There are currently 61 CDx on the market. In 2013, there were three
is faster than the ability to
drugs that were approved with companion diagnostics in their
generate and assess
labeling. Experts have said that there will be three to five CDx
biomarkers relevant to
approvals per year over the next five years (Mullard 2014a). Existing
the drug (Woodcock
CDx have focused primarily on oncology, but the next wave will be
2010). Thus, it has been a
autoimmune diseases and inflammation, as there are biomarkers
major challenge to
readily available (Opar 2011).
integrate CDx
FDA has provided concept papers and draft guidance
development into drug
documents on how best to integrate CDx into drug development in a
development. Developing
clear and cohesive manner. However, industry has pushed back
a CDx also requires
against these documents, calling them unrealistic guidelines that will
collaboration within and
hinder biomedical innovation (Ray 2013). Thus far, regulation has
across stakeholder
been done on a case-by-case basis within the context of the specific
groups. For example,
drug (Dennis et al. 2013; Woodcock et al. 2011; FDA 2014). The FDA
within the FDA, CDRH and
has stated that they are trying to be as flexible as possible because
CDER must play a
they are still learning, even though flexibility doesn't lend itself to
collaborative regulatory
defining a predictable pathway (Mansfield 2014). Industry has
role. Regulatory
complained that the regulations have been inconsistent-in some
collaborations also need
cases the drug and CDx must be approved together, in other cases
to occur on a global scale.
they are not, and finally, in certain cases, the CDx can be an LDT as
Currently, CDx regulations
opposed to a FDA-approved IVD (Carver 2010), without clear criteria
are very different at the
for distinction.
EMA; in the EU, a broad
range of CDx are allowed
Box 1. The importance of CDx and the controversy between industry
and there are no EMAand the FDA.
approval requirements
before a company can market a CDx (Pignatti et al. 2014). Within industry, closer collaborations need to
occur between diagnostic companies and pharmaceutical companies. These companies need to develop
innovative methods for pooling their data and agreeing upon analytical methods. Experts have also
proposed ways in which the companies can resolve IP issues in the co-development process (Keeling &
Roth 2008). Across stakeholder groups, the FDA and industry need to further collaborate as well and
resolve the debate around CDx (Box 1).
Tingyu Liu 120
The CDx landscape is increasing in complexity and requires regulation. For example, there are at
least ten HER2 tests that are FDA-approved CDx, which complicate assay selection (Parkinson et al.
2014). In addition, despite industry backlash, the FDA has repeatedly expressed its intent to regulate all
CDx, even if they are LDTs (Trusheim et al. 2013). This is a decision with widespread ramifications
because there is increasing evidence as we move into the realm of more personalized treatments that
some kind of CDx will be a part of all (or most) drug approvals (Drucker & Krapfenbauer 2013).
2.4: Summary: The Importance of MSCs in Biomarker Adoption
The path from an exploratory biomarker to a biomarker qualified for a specific application context can
be long and unpredictable (Vaidya & Bonventre 2010). The main gaps are summarized in Table 5. Note
that these gaps are not mutually exclusive. For example, a biomarker used in drug development can
later become a CDx, and thus R&D evidentiary standards issues also becomes inherent in CDx.
Table 5. Gaps and challenges preventing biomarker adoption.
Co-development is difficult to achieve in practice
No high level of cross-stakeholder collaboration; e.g. pharmaceutical and
diagnostic companies are not used to working together
Lack of global harmonization; FDA and EMA have distinctly different
regulations
No formal regulatory pathway & difficulty aligning stakeholders towards an
accepted pathway
Drug-IP conflicts limit sharing of knowledge & data
Rapid pace of new data and knowledge generation
Difficulty demonstrating analytical validity at small companies or acad(
researchers
Unclear how to assess which biomarker is the most robust
Biological variability and heterogeneity of patient samples
Need for multiple BM signatures for disease detection and monitoring
Poor standards on patient selection & sample collecting and handling
Stakeholders work in functional silos with limited communication
Tingyu Liul21
It is clear from this section that the main theme underlying current challenges in biomarker
adoption is the misalignment, misinformation, and misunderstanding between the different stakeholder
groups. Stakeholders work in functional siloes and cross-stakeholder awareness is limited, leading to
bottlenecks that prevent the movement of biomarker discoveries through increasing stages of maturity
(Keeling & Roth 2008). The lack of global regulatory harmonization policies on CDx regulations is one
pertinent example (Pignatti et al. 2014). Increased dialogue and understanding between stakeholders
will help shed light on relevant issues and lead to agreeable resolutions.
The level of resources needed (data, samples, expertise, financial, etc.) to discover, develop, and
evaluate biomarkers makes this field particularly fertile for collaboration and intervention among key
stakeholder groups. MSCs are able to tackle large undertakings such as generating and harnessing highquality evidence to demonstrate clinical utility or to maintain biomarker databases as more data is
produced (Parkinson et al. 2014; Zineh & Lesko 2009). The work and impact of leading cross-stakeholder
collaborations in addressing the main gaps listed in Table 5 are discussed in the next sections.
Tingyu Liu 122
Section 3: The impact of MSCs in promoting biomarker adoption in drug
R&D
The first regulatory pathway to promote biomarker adoption in R&D was officially introduced in the
form of a guidance document released by the FDA and EMA (FDA 2014). The document described the
qualification process for drug development tools (DDTs), which includes biomarkers, clinical outcome
assessments (COAs), and animal models. These tools are intended to be used in multiple drug
development programs to streamline the R&D process and increase the odds for success. For the
purposes of this thesis, the focus will be on the biomarker aspect of the DDT and the specific framework
for the biomarker qualification process (BQP) of the guidance.9
Before the BQP was established, any biomarker used in drug development would be evaluated
by regulatory agencies on a case-by-case basis which was both sponsor- and drug-specific. This means
that the sponsor will only perform the tasks necessary to develop the biomarker for a specific drugeven if the biomarker can be applied more widely. Also, in these cases, the information remains
proprietary (Dennis et al. 2013; Woodcock et al. 2011; FDA 2014). The BQP aims to establish a global,
rather than product-specific, fitness for use (Woodcock 2010).
The BQP represents the first concerted effort to address some of the gaps described in Section 2
(Table 5) by providing a regulatory pathway for biomarker adoption to aid in drug development
(Woodcock 2010). The BQP was conceived and developed in a multi-stakeholder environment, with
close collaborations between academics, industry, and regulators both in the US and EU. The BQP is one
of the main outputs of FDA's 2004 Critical Path Initiative (CPI), which was launched to foster the
development of new evaluation tools and drug trial standards by focusing on pre-competitive
collaborations and sharing of data (Yu 2014; Mahajan & Gupta 2010; Kelloff & Sigman 2012). Indeed, the
guidance itself specifically encourages collaborations since "substantial effort is involved in achieving
qualification" (FDA 2014).
Qualifying a biomarker is seen as a lengthy, resource-intensive process that an individual drug
company usually cannot accomplish on its own. As BQP participants have stated, a qualified biomarker is
"likely most appropriate for roadblocks in particular therapeutic areas or aspects of drug development
where there is pressing public health need and where cooperation between drug companies and other
entities is imperative for advancement" (Dennis et al. 2013). Safety biomarkers represent a good
example. For instance, drug-induced liver injury is the leading cause of drug failures in clinical
development and market withdrawals (Chen et al. 2011). Safety biomarkers are critically needed along
the healthcare value chain and can be used in many drug development programs. Qualifying safety
biomarkers is more likely to be feasible in a collaborative environment because they require a large
amount of evidence that would be difficult, if not impossible, to find within one drug company
(Woodcock 2010; Amur et al. 2008). The BQP allows stakeholders to work together in new ways-not
just between industry and regulatory agencies, but also between preclinical scientists and clinicians-to
share expertise and identify and address key research needs (Dennis et al. 2013). Indeed, regulatory
9
The framework can be broadly described as:
1) Initiation: sponsor submits a letter of intent, which is deemed acceptable by CDER.
2) Consultation and advice: an initial briefing package is submitted, iterative meetings and correspondence
occurs to review the submission, highlight gaps, and review any additional data that is submitted to
address those gaps.
3) Review of the full qualification package: members of the review team offer their assessment and a
combined executive summary is published for public comment before CDER issues the final qualification
recommendation.
Tingyu Liu 123
authorities have noted that over 80% of BQP submissions have been from consortia (Woodcock et al.
2011).
According to the FDA guidance, a biomarker that is qualified as a DDT signifies that "within the
stated context of use (COU), the DDT can be relied on to have a specific interpretation and application in
drug development and regulatory review. Once a DDT has been qualified for a specific COU in drug
development, it can be used to produce analytically valid measurements. The DDT...can be used by drug
developers...without the relevant CDER review group reconsidering and reconfirming the suitability of
the DDT" (FDA 2014). There are a few key points to be noted from this definition.
First, the COU describes the circumstances in which the biomarker can be used, based on the
available supporting data (FDA 2014). The COU was mentioned a few times in the Section 2 discussion
around evidentiary standards, because it is imperative to understand that the supporting data is only
valid within a specific COU. Thus, the data must be reevaluated if a biomarker is used in a different way,
or in another medical context (Altar et al. 2008).
Second, the biomarker "can be used to produce analytically valid measurements." Again, it is
important to note that the analytical validity of a biomarker depends on its COU; certain contexts might
only require an analytical range that is a subset of the whole available range (Campion et al. 2013).
Qualification prespecifies that a biomarker has demonstrated adequate reliability, sensitivity, and
specificity for a specific context of use (Stephenson et al. 2013). Although analytical validity is a term
that generally encompasses both the biomarker and its measurement device, this DDT guidance
specifies that the biomarker is "conceptually independent" of its device. This distinction is due to the
fact that the guidance is written by CDER, who has no authority over devices. Thus, evaluation of the
device must be considered separately from biomarker qualification and the BQP uses devices that have
already been evaluated (Woodcock et al. 2011).
Finally, the biomarker can be used without reconsidering and reconfirming its suitability. The
biomarker may be used across multiple clinical disorders, drugs, or drug classes. This would save the
resources of both the regulatory agencies and the drug companies. Many experts hope that the BQP will
result in a database of robust biomarkers that have been vetted by regulatory authorities, which will
help streamline drug R&D and mitigate the high attrition rates seen in clinical trials (Dennis et al. 2013).
In addition, if another sponsor wants to expand a biomarker's qualified COU, it can build upon the
available knowledge, as it is not proprietary.
The development of the BQP highlights the importance of collaboration and bold policy changes
in healthcare innovation. The BQP has the potential to lead to a harmonized and trusted pathway for
biomarker use. The emergence of the process and its success would not have been possible without the
work of various MSCs involving the regulatory agencies, especially the MSCs that were initiated by the
CPI. These organizations contributed to the various policies in the form of regulatory guidance
documents to address some of the challenges of biomarker adoption discussed in Section 2 (Table 5).
Section 3.1: Prequel: Voluntary eXploratory Data Submissions (VXDS) at the
regulatory agencies as an example of collaborations across stakeholders at the
global scale
The genesis of the BQP is the VXDS program, a collaborative program initiated by the FDA to explore the
use of pharmacogenomic data in R&D (Vaidya & Bonventre 2010). The program revealed to the
stakeholders that a regulatory path from exploratory biomarkers to biomarkers qualified for a specific
context of use was sorely needed.
The human genome was sequenced in 2000, leading to the discovery of many pharmacogenetic
biomarkers (Goodsaid & Papaluca 2010). Such biomarkers were increasingly being used in preclinical
drug development (Goodsaid & Frueh 2006), but the FDA was not sure how to handle this influx of novel
Tingyu Liu 124
pharmacogenomic data. In an effort to promote regulatory science in the light of genomic progress, the
FDA initiated meetings with industry to discuss appropriate applications of such data. At the first
meeting in May 2002, the FDA introduced a novel, voluntary submission process for pharmacogenomics
data, called Voluntary Genomic Data Submission (VGDS). VGDS allows sponsors to submit claims for
qualification through informal meetings with regulatory agencies (Orr et al. 2007). It was envisioned that
these informal meetings would occur under the concept of "Safe Harbor" in which the submissions were
evaluated under an "exempt status" so they would not compromise regular drug submissions. In this
manner, the stakeholders developed a platform for an open, scientifically-driven information exchange
between the FDA and industry so that both parties could learn how to analyze and integrate biomarker
data and enable its use (Frueh et al. 2006).
The FDA used what they learned from the first workshop to draft "Guidance for Industry:
Pharmacogenomic Data Submissions" and held a second workshop immediately after to discuss this
draft with industry in November 2003 (Frueh et al. 2006). The draft guidance formally introduced the
VGDS process along with a new Interdisciplinary Pharmacogenomic Review Group (IPRG) charged with
reviewing the submissions (Goodsaid & Frueh 2007). IPRG was FDA-wide and involved CDER, CDRH, and
CBER collaborations to help resolve the issues surrounding analytical evaluation of new genomic-based
biomarkers and their evidentiary standards (Gutman & Kessler 2006; Goodsaid & Frueh 2007). The IPRG
had the power to create a review team (termed the Biomarker Qualification Review Team, BQRT) with
the right scientific expertise to assess biomarker datasets.
Also in 2003, FDA and EMA authorities held their first joint discussions with sponsors on VGDS
packages under the framework of a bilateral confidentiality agreement (Goodsaid & Papaluca 2010).
Due to the success of this meeting, the VGDS was expanded, with an option for sponsors to have joint
FDA-EMA VGDS briefing meetings. In 2006, the EMA and FDA issued the "Guiding Principles for
Processing Joint FDA/EMA VGDSs" (Orr et al. 2007). EMA also released its own guideline for industry on
how to prepare for dialogue and submissions with the Pharmacogenetics Working Party (EMA 2006).
When the FDA released its final guidance document, it held a third workshop to discuss the issue
of varying data quality and standards in submissions (Frueh et al. 2006). The group also decided to
change the phrase VGDS to VXDS, or Voluntary eXploratory Data Submissions, because they wanted to
formally recognize that submissions extended beyond genomic data and also included AD and
depression biomarkers, as well as other "-omics" type exploratory technologies (Goodsaid & Frueh
2007; Orr et al. 2007).
Figure 2. The timeline of voluntary data submissions at the FDA, adapted from Orr et al., 2007.
Tingyu Liu 125
The impact of the VXDS process (Figure 2) was significant. Data sharing helped educate both the
FDA and industry and increased the overall knowledge base (Goodsaid & Frueh 2006). The FDA learned
how to analyze emerging data and innovative approaches. Industry also received feedback on trial
designs, methodologies, and data interpretation. For example, during one of the first submissions, the
FDA and the sponsor realized that they used different normalization and statistical approaches to
analyze the data, which resulted in different interpretations (Goodsaid et al. 2010). Several sponsors
have used these types of informal interactions with the FDA as a stepping stone to present the same or
related data later in a regulatory context to streamline drug submissions, update a drug label, or
construct novel adaptive phase III clinical trial designs (Orr et al. 2007).
The collaborative nature of this program led to new policies on genomic data in the form of
regulatory guidance documents that both FDA and industry found agreeable, leading to discussions on
what new regulations might be needed for the future. For example, AstraZeneca's VXDS submission led
the team to the idea that conditional approval may be needed in the co-development of safety
biomarkers to allow extensive monitoring of both the drug and biomarker after they had been launched
(Goodsaid et al. 2010).
There have been more than fifty VGDS/VXDS submissions since the program's inception
(Goodsaid & Mattes 2013). Approximately two-thirds of the submissions have focused on clinical
applications of exploratory biomarkers (Goodsaid et al. 2010). However, the VXDS program seems to
have outlived its original purpose, as there were no updates beyond 2010 (FDA 2010), although the EMA
counterpart continued to hold meetings in 2014 (EMA 2014).
Participants have discussed ways of expanding the scope and goals of the VXDS. For example,
regulatory authorities have discussed the possibility that they might serve as a broker to strategically
bring together multiple parties to provide different perspectives on biomarkers in various contexts,
which could lead to fruitful collaborations (Goodsaid et al. 2010). Indeed, a form of this vision did come
to pass: multiple pharmaceutical companies collaborated with the guidance of the FDA to advance
biomarker adoption via the Biomarker Qualification Program.
Section 3.2: The Biomarker Qualification Program
The evolution of the VXDS program into the Biomarker Qualification Program is an exemplary story that
demonstrates how FDA's launch of CPI and its resulting precompetitive initiatives0 pushed
collaborations to the next level to promote innovations in regulatory science. Regulators recognized that
while VXDS remains a valuable pathway, "there is a need...to provide a framework for the development
and regulatory acceptance of scientific tools for use in drug development programs [which has] led to
the establishment of a pilot project for biomarker qualification" (Goodsaid & Mattes 2013).
The initial process for safety biomarker qualification was done through industry collaboration in
a regulatory pathway structure exercise. The collaborative boundaries were pushed further while crossvalidating the safety biomarkers with other industry stakeholders who were members of the Critical
Path Institute's Predictive Safety Testing Consortium (PSTC). The collaborative regulatory pathway
structure exercise expanded upon a VXDS submission: the FDA and Novartis initiated a 2-year
cooperative R&D agreement (CRADA) to design a process map for qualifying preclinical safety
1 Note that CPI led to the formation of the Critical Path Institute (C-Path), which was created with support and
funding from the FDA, Science Foundation Arizona, and the Tucson community. C-Path is a non-profit organization
that serves as a neutral third party to manage funding and protect IP for its MSCs. Currently, there are six C-Path
MSCs: Coalition Against Major Diseases (CAMD), Coalition for Accelerating Standards & Therapies (CFAST), Critical
Path to TB Drug Regimens (CPTR), Multiple Sclerosis Outcome Assessments Consortium (MSOAC), Polycystic
Kidney Disease Outcome Consortium (PKD), Patient-Reported Outcome Consortium (PRO Consortium), and
Predictive Safety Testing Consortium (PSTC).
Tingyu Liu 126
biomarkers and testing this pilot process by submitting the biomarker data to the FDA for qualification
as Probably Valid Biomarkers (see Glossary A). CRADAs are important tools to promote innovations in
regulatory science because they allow the FDA to leverage industry's resources (funds, data, facilities)
while industry is able to develop a close working relationship with the FDA and obtain valuable
consultations and advice (Clinton & Wechsler 2006). To cross-validate the biomarkers, Novartis' data
was combined with data from other pharmaceutical companies who were PSTC members so that the
biomarkers could be qualified to achieve Known Valid Biomarkers status (Goodsaid & Frueh 2007;
Marrer & Dieterle 2007).
PSTC's final package of nine kidney safety biomarkers for preclinical use to describe
nephrotoxicity in rats was submitted to the FDA and EMA for qualification in June 2007; seven were
qualified a year later (Goodsaid et al. 2008). In light of this success, the Biomarker Qualification Program
was officially introduced in 2009 (Goodsaid & Mattes 2013). Since then, a number of qualifications have
occurred (Figure 3, bottom).
The evolution from VXDS to the BQP illustrates how industry has answered regulators' call and
helped them enhance regulatory science. This evolution led to new regulatory policies that have helped
clarify some of the evidentiary standards around biomarker use. The knowledge that regulatory
agencies gained from these collaborations and qualifications has allowed them to publish several
guidance documents which serve to share their current thinking around biomarker policy with industry
(Figure 3, top).
JEA released draft
guidance on codevelolxnentof
o
Drun
assays in drug
~~developrnentI
40
......
released draft
guideline on
biomarkers in clinical
development&
patient selection
EA
eJ
E
.2
Figure 3. The timeline of the BQP at the FDA and EMA, including relevant tangible outputs of
regulatory policies and qualified biomarkers. The MSCs that have submitted qualified biomarkers
include: PSTC, CAMD, HESI, and ADNI. Additional information on these biomarkers can be found in
Appendix A.
Tingyu Liu 127
Section 3.3: Analyzing key BQP outputs
To enhance the understanding of collaborations in biomarker adoption, we developed a framework to
analyze the contributions of the MSCs across four areas that involve important translational enablers for
biomarker adoption. Starting from the scientific challenges addressed by key MSCs, we expand to
describe their approaches and achievements on novel processes and policies toward an increased
systems capacity for therapeutic innovation through biomarker adoption. The four areas can be
described as: 1) creating data and data standards, 2) developing new scientific processes, 3)
operationalizing and organizing innovative collaborations, 4) delivering tangible outputs (Table 6).
(1) The biggest data contribution is simply the fact that industry shared their proprietary data to
be analyzed in a collaborative setting. This necessitated the creation of a standardized data format
which is vital for broad-based data harmonization and knowledge sharing. For example, CAMD
collaborated with Clinical Data Interchange Standards Consortium (CDISC) to publish the first data
standard for AD, which led to the first database of combined clinical trials to be openly shared by
companies with qualified researchers worldwide. The FDA has recognized the importance of such efforts
in streamlining the R&D process and mandated that CDISC's clinical standards will be required for
submissions by 2017 (Stephenson 2013). In addition, CAMD used this data in their AD biomarker
qualification package (EMA). However, more work still needs to be done. For example, during the
qualification of the first biomarker package, PSTC mentioned that a common database and data
submission format for nonclinical and clinical biomarker data was not available; the group had been
using Excel, but they stressed that it was a suboptimal platform for such processes (Dieterle et al. 2010).
Thus, it is unclear if the different MSCs involved in the qualification process are using the same data
formats and platforms.
(2) The main scientific process introduced by these MSCs is, of course, the BQP. According to the
FDA, they have already received many drug submissions that make use of biomarkers qualified by the
BQP (Goodsaid & Mendrick 2010). In addition, the MSCs contribute to the scientific process of the BQP
by providing "detailed documentation of the review process, to provide guidance for future regulatory
submissions" (Dieterle et al. 2010). For example, PSTC publishes all biomarker research, even if the
qualification submission is not attained (Dennis et al. 2013). PSTC also explicitly introduced the
"Progressive Qualification Framework" which keeps dialogue and evidence gathering open so regulatory
agencies can anticipate additional data and evaluations in the future (Sistare et al. 2010). This would
include data and analysis from any clinical studies of the currently qualified biomarkers. However, there
is no guarantee that sponsors will update and curate the qualified biomarker's database as new
evidence is generated (Walker et al. 2013).
The BQP has also stimulated similar scientific processes outside of CDER. Recently, CDER used
what it has learned from the BQP to help CDRH frame a similar DDT guidance entitled Medical Device
Development Tools (MDDT). CDRH can reach out to CDER to obtain advice about implementing such
tools. Like the DDT, the MDDT guidance also places emphases on MSC contributions (FDA 2013a).
Indeed, the Medical Device Innovation Consortium (MDIC) was launched in partnership with the FDA
with the goal of advancing medical device regulatory science (Anon 2013).
(3) The collaborative structure that the MSCs have agreed upon allowed them to share and work
with precompetitive data with a mutually valuable end result. PSTC, as the first MSC to submit a
biomarker package to the BQP, developed a "unique Consortium Agreement" that defined the
operational and organizational issues around the collaborative structure, including anti-trust issues,
confidentiality guarantees, and so forth (Goodsaid et al. 2008; Mattes et al. 2010).
Moreover, innovative collaborations would foster biomarker adoption by structuring a
streamlined qualification process between FDA and EMA. PSTC's project on renal safety biomarkers has
demonstrated the early impact of streamlining a qualification process at the global level. Although both
Tingyu Liu 128
FDA and EMA have similar qualification processes, they only share and discuss biomarkers that are
jointly submitted (Walker et al. 2013; NCI 2012). Yet industry have stated that joint submissions are
challenging in practice (Manolis et al. 2011). One of the key challenges towards global harmonization is
the difference in qualification opinions. For example, unlike the FDA, the EMA is open to qualifying a
biomarker for certain COU even in the absence of consensus in the field on standardization and cut-off
values, with the understanding that the guidance can be updated as new data are obtained. CAMD is
currently working to align the qualification process to ensure that data submitted to one regulatory
body would also meet the requirements of another agency, although the details of this plan is still
unclear (Stephenson et al. 2013).
(4) The BQP stimulated the publication of many guidance documents that outline innovative
policies for biomarker adoption (Figure 3, top). However, for BQP to widen its impact, the guidance
documents need to be expanded upon. For example, there is no guidance for how to apply retrospective
studies to regulatory decision making (Khleif et al. 2010). Cardiac troponins was the only qualified
biomarker whose submission was entirely based on retrospective data (Hausner et al. 2013). The FDA
has also recognized that they may need to think about how they would handle biosignature qualification
(more than 1 biomarker) submissions (FDA 2014). Some experts have proposed an updated document
that provides more guidance around expected evidentiary standards (Campion et al. 2013). Explicit
guidance should secure inclusion of stakeholder views. For example, some experts have voiced their
concern that the BQP team at the FDA has moved towards a more narrow context of use and a
conservative interpretation of the data, and the process is becoming more time-consuming and
resource intensive (Dennis et al. 2013). As another example, in the 2012 NCI-FDA-NIST workshop report,
a regulator involved in the process stated that often, the FDA and the submitter get stuck in an "infinite
loop" during the consultation and advice stage of the BQP and he suggested abolishing this inefficient
aspect of the process (NCI 2012).
Table 6. MSC contributions for biomarker adoption in R&D.
Tingyu Liu 129
Section 3.4: Conclusion and next steps
The evolution of the VXDS program and the BQP are key historical examples that illustrate the impact of
MSCs in promoting novel pathways for biomarker adoption in drug R&D. Experts in the field have
discussed where the BQP should go next. One generally unanimous next step envisioned by the MSCs is
progressing the qualified preclinical biomarkers into clinical qualification. Industry has already voiced its
support for such a program, because currently, sponsors need to discuss their plans with regulatory
groups on a case-by-case basis if they wish to use these qualified biomarkers in clinical studies (Campion
et al. 2013). PSTC have already begun clinical studies and have initiated collaborations with more
clinically-focused MSCs such as the Biomarkers Consortium and the SAFE-T consortium (Dennis et al.
2013). Once qualified clinical biomarkers exist, regulatory agencies can more fully explore how the BQP
addresses issues of clinical validity and utility, thus widening the scope of the BQP in the healthcare
value chain.
The key challenge with the current scientific process is that the pathway is slow and complex,
which does not incentivize industry to contribute their resources to further the BQP. Industry also finds
it difficult to integrate biomarker qualification with drug development when the former process is too
slow and vague to be of significant use (Campion et al. 2013). Defining the level of assay validation
required for each COU and investing in assay development and validation is tricky, and there are no
specific guidance documents for this issue (Dennis et al. 2013).
Although MSCs have contributed significantly to the establishment of the BQP, the utility of the
BQP alone is limited due to its narrow scope. In the next section, the work of other MSCs will be
illustrated. These MSCs possess broader goals and focus on biomarker adoption for patient care,
including biomarkers for later-stage clinical trials and diagnostics.
Tingyu Liul30
Section 4: MSCs' role in promoting biomarker adoption in later-stage
R&D and patient care
From our analysis so far, it has become evident that many processes or regulatory policies are needed to
address evidentiary standards for the different types of biomarkers and their COUs. Although the BQP
represents one important guideline, it is not the only one being developed. MSCs involved in the BQP
have thus far focused mainly on biomarkers for early-stage drug R&D. However, there are other MSCs
that are focusing on biomarkers involved in later-stage R&D and clinical care, to allow patients to fully
capitalize on the promises of personalized medicine. Biomarkers in the patient care setting cannot be
decoupled from their measurement devices, as was seen in the case in the BQP. The BQP evaluated a
biomarker in terms of its ability to be a tool for drug discovery, whereas a biomarker used in clinical
practice for diagnosis, prognosis, or disease management becomes a product per se.
Traditional IVDs encompass hematology, glucose-monitoring, and histology. Newer IVDs include
tests that measure DNA, RNA, and protein expression. There are different degrees of stringency in the
use of IVDs. In the case of a CDx such as the Herceptest for the drug Herceptin, the IVD is required for
the safe and efficacious use of the drug. An IVD can also be recommended in the use of a drug. For
example, individuals with low/moderate thiopurine S-methyltransferase (TPMT) activity are predisposed
to mercaptopurine toxicity and will show evidence of severe bone marrow toxicity; for such patients, a
TPMT assay is recommended. A drug's label can also contain information about changes in efficacy,
dosage, or toxicity in a subset of patients, but does not require or recommend the use of an IVD. And
finally, an IVD can be thought of as being relatively independent of any drug; for example, 23andMe
discusses genetic associations with various diseases, but does not point to any specific therapies.
MSCs in the IVD space encounter new challenging issues because 1) they need to deal with the
uncertain outcome of the debate on IVD regulatory pathways; 2) they may need to conduct large clinical
trials, which take significant resources but provide very few incentives; and 3) the line between
precompetitive and competitive collaboration is more blurry at this stage. This thesis focuses on two key
global MSCs that have distinct goals and have approached the particular challenges of biomarker
adoption in the patient care and management settings.
Section 4.1: The Biomarkers Consortium
The Biomarkers Consortium was launched in October 2006 (Box 2) with the goal of discovering,
developing, and qualifying biomarkers to support new drug development, preventive medicine, and
medical diagnostics. From the outset, they have explicitly stated that they are helping to create a new
era of personalized medicine, extending their goals beyond developing biomarkers for R&D by
addressing the critical later stages of diagnostics development. "The Biomarkers Consortium can help
advance evidence generation of clinical utility for molecular diagnostic tests with expanded access to
sources of funding and data that are needed to support well-designed studies that produce evidence of
clinical utility" (Parkinson et al. 2014). For example, the Cancer Steering Committee Clinical Utility
Working Group is helping to provide practical definitions of clinical utility and guidelines on the
information needed by payers to support reimbursement for molecular diagnostic assays used in cancer
therapy (Parkinson et al. 2014).
Tingyu Liu 131
The Biomarkers
Consortium has completed five
projects to date and currently has
six that are active. Their first
simptete rectv. w
st
adiponectin, demonstrating its
value as a robust predictor of
glycemic response to peroxisome
pliferat re-pctsevttedereceptor
proliferator-activated receptor
treatment of diabetes (Wagner et
al. 2009). This first project
contributed significantly to the
optimization of performance for
collaborative environments. Key
learnings include the importance
of having regular face-to-face
meetings and teleconferences,
and collaborative resources, such
as a website to share information
and project updates. The team
The Biomarkers Consortium was an initiative of the
Foundation of the National Institute of Health (FNIH), a
nonprofit that is associated with but independent of NIH. FNIH
is authorized by Congress to broker relationships between NIH
and industry, academia, and philanthropies, so that it can seek
funding from NIH institutes as well as private partners (IOM
2008). This sort of arrangement is advantageous to FNIHfostered partnerships like the Biomarkers Consortium because
it incentivizes industry participation; the basic investment
has
already been funded by NIH, and industry is more willing to
leverage a preexisting investment (IOM 2010b). The
Biomarkers Consortium is divided into a number of diseasespecific steering committees (cancer, inflammation and
immunity, metabolic disorders, and neuroscience) that are
also composed of multiple stakeholders from academia,
government, industry, and non-profit/advocacy organizations.
The committees solicit project proposals from all sources.
Although the Biomarkers Consortium have defined their
projects to be pre-competitive, issues revolving around IP and
data sharing are resolved on a case-by-case basis (IOM 2008).
also decided that the best
Box 2. Background on the Biomarkers Consortium.
method to handle legal matters
was to designate one legal liaison per company (Wagner et al. 2010).
Section 4.1.1: I-SPY 2 Trials
A key example of the Biomarkers Consortium's work is the I-SPY 2 trial (Investigation of Serial Studies to
Predict Your Therapeutic Response with Imaging And Molecular Analysis), a novel phase II adaptive trial
model to accelerate the approval process of effective drugs for breast cancer using biomarkers. I-SPY 2
leveraged two features which were established during I-SPY 1 (Box 3): the infrastructure to ensure
accurate and consistent data collection and sharing, and the use of pCR as a primary endpoint.
I-SPY 2 was launched in 2010 with a budget of $26 million and a timeline of five years (IOM
2010b). As a precompetitive adaptive screening trial, it is a pioneering example for both drug and
biomarker development. Up to twelve investigational drugs, paired with biomarkers that are predictive
of the drug response, will be tested (Box 4). The trial's adaptive design focuses on progressive, real-time
learning of which biomarker profiles predict response to each drug (Barker et al. 2009). The trial uses
Bayesian approaches to determine the predictive probability success in a phase Ill confirmatory trial
(Barker et al. 2009). Bayesian approaches are considered quite novel and have faced limited adoption in
drug development, so I-SPY 2 might increase its use (Allison 2010). Initially, randomization to new drug
regimens is balanced. As results become available, predictive indexes are built for each drug and
biomarker combination. At this point, randomization is adapted towards assigning new patients to the
drugs that are predicted to provide the highest benefit. After approximately sixty patients, drugs that do
not sufficiently benefit the patients with any biomarker profiles will be dropped from the study. This is a
key development because it is the first time that a trial can compare drug effectiveness in real time and
replace ineffective drugs with better candidates.
Biomarkers used in I-SPY 2 belong to three classes (Barker et al. 2009):
1) Standard biomarkers are already in clinical use or FDA-approved and will be used to
determine patient eligibility and randomization.
Tingyu Liu 132
2)
Qualifying biomarkers have not yet been FDA-approved and will be evaluated under
Clinical Laboratory Improvement Amendment (CLIA) quality standards. These
biomarkers are termed "hypothesis-testing" because they show promise for predicting
which patients will respond to which agents.
3) Exploratory biomarkers are also termed "hypothesis-generating"; they may possess
predictive or prognostic value for breast cancer treatment, but not much information is
known yet.
Note that these biomarkers are drug-specific, suggesting that after undergoing the necessary evaluation,
they will become either a CDx or a recommended IVD for use during the diagnostic process (L. J.
Esserman et al. 2012). Experts have stated that while the sojourn time for any drug in the trial is too
short to assess the validity of the biomarker, positive results could lead to FDA approval to use the
biomarker in a follow-up registration trial (IOM 2010b). Promising data can support a PMA or a
regulatory request to use a biomarker to stratify patients in a Phase Ill validation study (IOM 2010b).
It took significant time and effort
I-SPY 1 began in 2002. I-SPY 1 was not a Biomarkers
to plan, agree on, and coordinate the
Consortium project but a collaboration of the
unique aspects of I-SPY 2. Indeed,
American College of Radiology Imaging Network
developing the protocol alone took two(ACRIN), Cancer and Leukemia Group (CALGB), and the
and-a-half years, with many multiNational Cancer Institute (NCI)'s Specialized Program
stakeholder meetings involving the FDA.
of Research Excellence (SPORE). Its primary objective
The stakeholders that developed the Iwas to evaluate whether response to therapy, as
SPY 2 protocol set up the innovative
measured by MRI volume and pathologic complete
"master Investigational New Drug (IND)"
response (pCR), would predict recurrence-free survival
concept (Patlak 2010). Usually, multiple
(RFS). Its secondary objective was to build a resource
drugs and biomarkers require multiple
and establish standards for collected clinical,
trials, each with its own IND application.
molecular, genetic, and imaging biomarker data for a
However, for I-SPY 2, FNIH was chosen as
multicenter network that would support high-quality,
a trusted intermediary to hold the master
real-time biomarker evaluation for future trials of such
IND. Thus, when I-SPY 2 began, the IND
nature (L. Esserman et al. 2012). I-SPY 1 demonstrated
protocol had five potential therapies. As
that the predictive power of pCR is robust enough to
the trial progresses, other therapies will
be used as an early indicator of RFS, thus establishing
be submitted to the regulatory
pCR to be a surrogate endpoint (Esserman &
authorities for approval, but the master
Woodcock 2011). I-SPY 1 also set up standard
IND will only be modified with an
methods for collecting core biopsy material that are
additional appendix. This highlights the
then used in measuring and evaluating gene
importance of a neutral third party in
expression profiles, and for MRI-based tumor
successful collaborations.
evaluation (L. J. Esserman et al. 2012; Allison 2010).
The stakeholders have also
worked out an IP agreement that allows
Box 3. Background on I-SPY 1.
them to collaborate in a trial that extends
the boundaries of the pre-competitive line (Goldman et al. 2013). The participants have agreed that the
drug IP will remain with the company owning the IP, the biomarker IP will also remain within the
company but be licensed for use in the project, and any new IP will be managed by FNIH, which will
return royalties to the inventing organizations (IOM 2010b). However, it is unclear what new IP might be
created, and what kind of royalties would be implemented. This will set a strong precedent for
collaboration in the later stages of development, as collaboration is increasingly being pursued for
therapeutic innovation.
Tingyu Liu 133
The I-SPY 2 trial played an instrumental
role in the use of biomarkers to support drug
development. As a result of the learnings gained in
I-SPY 2, FDA published a draft guidance on using
pCR in neoadjuvant treatment for high-risk earlystage breast cancer; pCR would be used as a
surrogate endpoint to support accelerated
approval (Esserman et al. 2013; Prowell & Pazdur
2012). The guidance recognized for the first time
that neoadjuvant and adjuvant therapy resulted in
the same long-term outcomes, but the
neoadjuvant approach allows earlier measurement
of therapy response and provides information
about the prognosis of each patient. This draft
guidance addresses the lack of an approved
regulatory pathway for CDx.
Section 4.1.2: Analyzing the outputs of the
Cosoriumwould
Biomarkers Biomrkes
Consortium
The initial five drugs that entered the I-SPY 2
trial are: veliparib (ABT-888; Abbott),
conatumumab (AMG 655 & AMG 386;
Amgen), figitumumab, and neratinib (Pfizer,
Puma). In order to enter I-SPY 2, drugs must
have completed Phase I testing, and have
evidence of potential efficacy from preclinical
or clinical studies (Barker et al. 2009). I-SPY 2
just graduated its first two candidates:
AbbVie's veliparib & Puma Biotechnology's
neratinib (Fox 2014). Currently, the
Biomarkers Consortium is in discussion with
AbbVie and Puma about launching I-SPY 3, an
adaptive collaborative follow-on, although
the compounds might also go through
traditional company-sponsored trials. It
wol
e.inte
s ose how I
be interesting to see how IP issues
would be dealt with in the former case.
As in Section 3, the Biomarkers Consortium's
contributions to biomarker adoption can be broken Box 4. I-SPY 2 drugs and progress.
down into the four key areas of 1) creating data
and data standards, 2) developing new scientific processes, 3) operationalizing and organizing innovative
collaborations, and 4) delivering tangible outputs (Table 7).
(1) Like the MSCs mentioned in the previous section, the members of the Biomarkers
Consortium also shared precompetitive biomarker data that were aggregated and analyzed to produce
robust biomarkers. The I-SPY trials established data standards and data collection; industry stakeholders
demonstrated that their drugs' phase I trials were robust and shared their drug candidates to be used in
a phase II trial with common data formats.
(2) The scientific process of linking a biomarker to a drug through a trial like I-SPY 2 provides a
clearer pathway for the biomarker to become a diagnostic device, thus addressing the co-development
pathway gap discussed in Section 2 (Table 5). To streamline the process for subsequent trials, it is
imperative that the scientific process for determining the robustness of the drug and the requirements
for a biomarker to achieve exploratory, qualifying, or standard categorization are clearly established.
Secondly, the use of a Bayesian approach in drug trials is currently uncommon, but it is integral
for an adaptive trial. Thus, I-SPY 2 may lead to wider use and understanding of this scientific approach.
Finally, the I-SPY 2 trial provides a policy model for an adaptive, biomarker-driven screening trial with
multiple compounds. I-SPY 2 experts have discussed how this model can be used for other cancers that
possess emerging predictive markers and poor outcomes (L. J. Esserman et al. 2012). Similarly designed
trials such as BATTLE-2 and the Lung Cancer Master Protocol have already begun in the wake of I-SPY 2
(Fox 2014).
(3) The collaborative structure of the Biomarkers Consortium extends a step beyond MSCs such
as PSTC because stakeholders are working together on projects that tackle later-stage R&D biomarkers.
I-SPY 2 is especially groundbreaking as stakeholders are able to achieve collaboration even though the
trial has strong clinical and commercial implications. Such a concept may lead to the formation of a
collaborative I-SPY 3 trial. However, there has not been any public discussion of partnerships between
pharmaceutical companies and diagnostics companies following the results of I-SPY 2, although this
between-stakeholder collaboration is integral to the trial's ultimate success.
Tingyu Liu 134
(4) Most of the Biomarkers Consortium's completed projects so far have only resulted in
presentations and publications. Thus, it is not yet clear whether their work had concrete impact. The
main policies that have stemmed from the Biomarkers Consortium are FDA's draft guidance on pCR in
neoadjuvant treatment of high-risk early-stage breast cancer and the "master IND" procedure, which is
being adopted by other adaptive cancer trials. In addition, although the concept of I-SPY 2 can be seen
as an innovative, new regulatory pathway for CDx approval, it is still too early to appraise its success
since no drug or biomarker has been approved yet.
Table 7. The Biomarkers Consortium's contributions for biomarker adoption.
Section 4.2: The Early Detection Research Network (EDRN)
The EDRN focuses on accelerating the translation of biomarker information to clinical application and to
evaluate ways of testing cancer in its earliest stages. The EDRN was created by NCI in 1999, with the
initial goal of creating an informatics platform to connect different research networks, facilitate
collaboration, and promote efficiency and rigor in biomarker research. Originally, the different research
networks involved in biomarker development at NCI-including discovery/basic scientists, translational
scientists, clinical scientists, and data analysts-worked in relative isolation in their respective siloes. The
EDRN was developed to address the resulting disconnect between the abundance of biomarker
discoveries to final product output.
The EDRN is composed mainly of research laboratories where biomarker studies are conducted
and governed by a Steering Committee of Principal Investigators (Pis) and NCI staff. A Network
Consulting Team of external experts from different stakeholder groups reviews progress and
recommends new research opportunities. Moreover, the EDRN's Prioritization Subcommittee
establishes procedures for allocating resources and prioritizing research, including the establishment of
metrics for evaluating progress.
Tingyu Liu 135
The EDRN collaborates with organizations such as the Jet Propulsion Laboratory (JPL), to develop
informatics and standards, and discusses diagnostics requirements at biennial meetings with the FDA
and CMS. However, although the EDRN is part of the NCI, it is unclear if the EDRN share practices with
other NCI collaborations such as the Cancer Biomarker Collaborative (CBC), the Clinical Assay
Development Program (CADP) and the Molecular Characterization Clinical Assay Development
Laboratory (MC-CADL) (Williams et al. 2012).
The EDRN is composed mainly of scientists who, unlike traditional academic researchers, have a
broad understanding of the rigor and sample size necessary to move a biomarker through development.
The EDRN has built a rigorous and widely-known method of biomarker development and evaluation,
which serves as an informative example for MSCs in the field. The head of the EDRN has commented
that the "EDRN is one of the first organizations to recognize the importance of standardized, prospective
collection of biologic samples in the context of clinically relevant circumstances" (Srivastava 2013).
Table 8 illustrates the EDRN's 5-phase biomarker development guideline (Wagner & Srivastava
2012; Pepe et al. 2001). The EDRN notes that "there is no similar collaborative group that is capable of
...research from basic discovery to clinical validation" (Wagner & Srivastava 2012). The EDRN's large
reference laboratory and the EDRN Knowledge Environment (EKE) allow it to carry out rigorous
analytical and clinical validation studies. The EDRN has over 300,000 specimens and ten standard
reference specimen sets from standardized, well-characterized, matched cases and controls. EKE is the
informatics infrastructure that connects information about biomarkers, studies, specimens, and
resulting scientific data (Crichton et al. 2010). The EDRN used JPL's data system as a model to implement
their biomarker data analysis and research centers, demonstrating successful technology infusion and
agency transfer.
Table 8. The EDRN's 5-phase guideline of biomarker development.
Identify potentially useful biomarkers
Determine the capacity of biomarkers to distinguish between patients with cancer
and those without cancer (most biomarker don't progress beyond this phase)
Determine how well biomarkers detect preclinical disease by testing the markers
against specimens collected longitudinally from research cohorts
Identify the extent and characteristics of disease detected by the test and
determine the false referral rate
Evaluate both the role of biomarkers for detection of cancer and the overall
impact of screening on the population through large-scale population studies
With these processes and datasets in place, the EDRN is a pivotal case in the adoption of a
"vertical approach" to biomarker research. EDRN is divided into four main functions (Pepe et al. 2001):
1) The Biomarker Developmental Laboratories (BDL) develop and characterize new
biomarkers or refine existing biomarkers;
2) The Biomarker Reference Laboratories (BRL) analytically validate biomarkers using
blinded specimens, including technical development, refinement, and quality control;
3)
Clinical Validation Centers (CVC) clinically validate biomarkers by conducting a multi-site
biomarker validation trial; and
4) The Data Management and Coordinating Center (DMCC) performs statistical analysis
and supports the informatics infrastructure.
These laboratories and centers work together to move a biomarker along the 5-phase process described
in Table 8. Further, the DMCC team designed a process for phase 2/3 biomarker validation trials termed
Tingyu Liu 136
PRoBE (Prospective-specimen-collection-Retrospective-Blinded-Evaluation) in which specimens are
collected before diagnosis to determine whether the biomarker's performance characteristics are
sufficient for clinical translation (Prensner et al. 2012). This concept is not unique to the EDRN, although
it may possess a more specific, well-validated protocol for such purposes.
It is important to note that out of the 5-phase process, the EDRN only focuses on phases 1-3
(Table 8). Phases 4 and 5 require collaboration with the clinical trial community (Prensner et al. 2012).
This demonstrates that progressive involvement of stakeholders is needed to move a biomarker to full
clinical validation. It is not clear how smooth the transition is from phase 3 to 4. In addition, the EDRN
must engage with industry to license the biomarker and take it through FDA approval. In this case,
"industry" typically refers to diagnostic companies. The EDRN holds biannual industry forums that bring
together academic investigators and industry representatives interested in commercializing cancer
biomarkers (Wagner & Srivastava 2012). The EDRN views the practice of only engaging industry during
the latter half of the biomarker development process as its core strength. As the EDRN has stated,
"Without the pressure of venture capital, 'best science' rather than 'best business' practice permits
development of novel concepts that otherwise, because of commercial needs, might never reach
fruition" (Wagner & Srivastava 2012).
Section 4.2.1: Analyzing the EDRN outputs
Again, as we have seen, the EDRN's key areas of contribution to biomarker adoption are listed in Table 9
as 1) creating data and data standards, 2) developing new scientific processes, 3) operationalizing and
organizing innovative collaborations, and 4) delivering tangible outputs.
(1) In terms of data and standards, the EDRN stresses the maintenance of robust and wellcurated EKE and reference sets, which is a key foundation for successful biomarker development. In
addition, introducing the concept of rigorous standards and uniform data sets to research scientists
allows them to understand the magnitude of efforts necessary for biomarker adoption. The limitation of
the EDRN's data and standards is that their focus on developing early detection biomarkers may be too
narrow. Such a structured methodology can be applied beyond this limited category of biomarkers.
Thus, it may be more broadly beneficial if the EDRN could reach out to other biomarker collaborations
or companies and provide guidance on how to improve their systems.
(2) As with data and standards structures, the EDRN's process of biomarker discovery and
development also focuses on robust and rigorous processes. The EDRN's 5-phase approach and PRoBE
design can likewise be more broadly applied to biomarkers beyond early detection, to enable wider
learnings in the biomarker field. In addition, the EDRN's documents do not explicitly discuss their
scientific process in relation to the evidentiary standards framework of "analytical validity, clinical
validity, and clinical utility." Hence, the EDRN's scientific process may add to the confusion highlighted in
Section 2, where stakeholders use multiple frameworks and terminologies to describe similar concepts.
The EDRN has stated that its methods are "widely used." However, it is difficult for any organization to
validate such a statement.
(3) The EDRN often acknowledges the importance of collaborations. Indeed, their collaboration
with JPL allowed them to apply learnings to build their own informatics system. The EDRN was also one
of the seminal organizations that educated siloed scientific researchers on the importance of a broad
understanding of the whole value chain in biomarker discovery and development, including how to work
together in this space. In addition, although the EDRN typically does not engage industry until a
biomarker reaches phase 4 (Table 8), it understands that engagement is necessary because it does not
have the capability or resources to commercialize a biomarker. EDRN also prides itself on openness; it
has stated that new members constantly join the organization. The Associate Membership option is an
easy way to allow a new member access to the EDRN's reference sets for evaluating its own biomarkers.
More than two hundred collaborative projects have been initiated, and some have resulted in public-
Tingyu Liu 137
private partnerships (Srivastava 2013). Currently, the EDRN has more than fifteen collaborations with
biotechnology and diagnostics companies (Srivastava 2013).
(4) The EDRN has identified over 1000 biomarkers, prioritized 300, and completed 10 validation
trials. These biomarkers begin as CLIA-certified LDTs, but once enough samples have been obtained, the
end goal is FDA approval. Thus far, the EDRN has five FDA-approved biomarkers for various clinical
endpoints (Anderson & Kodukula 2013). As an example, one of the approved biomarkers was licensed to
Vermillion as the OVAl test. OVA1 was endorsed by the American Congress of Obstetricians and
Gynecologists and the product revenue in 2013 was $2.6 M (PRNewswire 2013; Vermillion 2013).
However, one of the challenges that the EDRN faces is the ability to track these commercial metrics;
industry may not wish to release data on how well an EDRN-licensed diagnostic is performing in the
marketplace. But such information is vital because it allows stakeholders to recognize the EDRN's
success and further its mission.
Although the EDRN's efforts have not led to explicit regulations or policies, its rigorous pathways
and fruitful collaborations can be seen as developing important policy tools. Collaborative research is a
cultural mindset, and it is one that is increasingly being adopted by researchers, in part due to
organizations such as the EDRN.
Table 9. EDRN's contributions for biomarker adoption.
Developed a 5-phase guideline for biomarker discovery and
development
Developed a process for phase 2/3 biomarker validation trials (PRoBE
design)
Enabled learning, evaluation, and optimization via process
communications and publications that describe both successes and
failures
5 FDA-approved diagnostics
5 diagnostics in CLIA-certified labs (pending future FDA-approval)
Tingyu Liu 138
Section 5: Lessons and recommendations to enhance the role of multistakeholder collaborations (MSCs) in biomarker development and
implementation
The last few decades have seen a gradual shift towards collaborations in an effort to generate, diffuse,
and utilize knowledge and innovation in science and technology. A significant body of research has
analyzed various distributed innovation systems, like the Triple Helix analytical framework that
conceptualizes university-industry-government interactions and identifies key gaps and opportunities
(Carlsson et al. 2002; Etzkowitz & Leydesdorff 1995; Ranga & Etzkowitz 2013).
On a similar ground, the present thesis has developed a proposed framework to map the unique
contributions of MSCs in the biomarker field across four areas that involve important translational
enablers, particularly data and data standards, novel scientific processes, key operational and
organizational aspects of collaborative innovation, as well as the prominent tangible outputs like novel
biomarkers and relevant regulatory guidelines. In this concluding section, we distill our analysis in each
of these areas into a general set of lessons and recommendations to highlight the role of MSCs in
advancing biomarker progress, and discuss current gaps, future challenges, and points of intervention.
Section 5.1: Creating data and data standards
Biomarkers comprise a particularly data and labor intensive field. The level of data sharing through
MSCs allows them to tackle projects that would not otherwise be feasible. The data needed for
qualification of nephrotoxicity biomarkers, for example, required contributions from multiple industry
members in PSTC (Goodsaid & Frueh 2007; Marrer & Dieterle 2007). On this ground, broad-based data
gathering and harmonization can support progressive standard setting in terms of formatting
requirements and quality control to facilitate subsequent stages of decision making (regulatory
submissions, designing novel clinical trials, etc.) This is what happened when ADNI began collecting MRI
scans, PET scans, and CSF samples from thousands of patients in hundreds of centers worldwide, and
devised common protocols to combine and compare the data (Miller 2009). Through this level of data
consolidation, ADNI has contributed substantially to recent revisions of diagnostic criteria for AD and
MCI (Aisen 2011). Moreover, other MSCs such as the Alzheimer's Association are building on ADNI's
methods, demonstrating increasing importance of the MSC in the field (Carrillo et al. 2013).
Data consolidation from various sources also serves to enhance understanding between
stakeholders and across stakeholder groups by decreasing uncertainty and increasing consistency in
their evidence interpretations and decision-making criteria. For example, the EDRN Knowledge
Environment (EKE) connects basic scientists with translational and clinical scientists, allowing these
stakeholders to develop a broader understanding of the magnitude of efforts necessary to develop
robust biomarkers (Crichton et al. 2010). In the VXDS program, industry shared data one-on-one with
regulators, revealing simple, yet crucial disconnects that could be remedied more easily (Orr et al. 2007;
Frueh et al. 2006). For example, during one of their meetings, the two stakeholder groups realized that
they used different normalization and statistical approaches which resulted in different interpretations
of the same data (Goodsaid et al. 2010). Several sponsors have used the VXDS program to gain a better
understanding of FDA's reasoning so they can streamline subsequent drug submissions, update a drug
label, or construct novel adaptive phase Ill clinical trial designs (Orr et al. 2007).
Section 5.2: Developing new scientific processes
The scientific processes developed by these MSCs are essential to break down large, complex bodies of
work into manageable workstreams. A characteristic example has been the definition of the COU
(context of use) by participants during the design of the DDT; explicitly defining the COU makes
Tingyu Liu
139
qualification more feasible (FDA 2014). Similarly, the EDRN's vertical approach moves biomarkers along
different well-defined workstreams with structured sets of criteria for assessment. The EDRN proactively
coordinates its resources so that there is a rapid and efficient "hand-off' of biomarkers and biomarkerrelevant information along the chain of laboratories and data centers (Wagner & Srivastava 2012).
The MSCs discussed in the previous sections have also helped establish new pathways for
biomarker development and use across R&D stages and disease areas. The BQP represents a
consolidated new path to gather diffused data and harness resources in a commercially uncertain space
by qualifying biomarkers for use in early-stage R&D. Moving into the clinical setting, the I-SPY 2 trial is
the first adaptive, biomarker-driven screening trial with multiple compounds for breast cancer. At this
stage, it is too early to determine the project's success, since the trial has not led to new therapies
and/or approved biomarkers as of this writing. The I-SPY 2 trial, and all subsequent efforts to be
informed, provide an important example of how MSCs can develop novel paths of knowledge and
outputs communication to increase our predictive capacity for key decisions around the deployment
and therapeutic impact of biomarkers across R&D stages and disease areas. The I-SPY model is already
being copied for other pressing diseases, such as lung cancer, in the BATTLE-2 trial and the Lung Cancer
Master Protocol (Fox 2014). Further, the Innovative Medicines Initiative (IMI), Europe's largest publicprivate initiative, recently announced an unprecedented E53 million ($73 million) call for proposals for
such a trial in Alzheimer's disease; at least twelve drug companies have already declared their
participation (Mullard 2014b). I-SPY 2 also represents a possible pathway for IVD development, if
successful (IOM 2010b). Similarly, EDRN's 5-phase process for early detection biomarkers represents a
rigorous, well-characterized pathway to bring diagnostics to market (Wagner & Srivastava 2012; Pepe et
al. 2001).
Section 5.3: Operationalizing and organizing innovative collaborations
As biomedical knowledge and science become more complex, specialized, and fragmented, it is
important to enhance the capacity for their integration across R&D functional siloes and stakeholder
groups. MSCs are important vehicles to pull together necessary resources for this convergence. They
allow industry to provide the necessary funds, data, and facilities while the FDA contributes advice and
guidelines for standards. Neutral parties such as nonprofits and academia offer an impartial space for
meetings and discussions. The importance of a neutral, trusted third party, such as FNIH, to hold the
master IND document for I-SPY 2 trials is also widely recognized (IOM 2010b). The very involvement in
an MSC provides stakeholders with a better understanding of what is needed to work together
effectively, including designing detailed IP agreements (IOM 2010b), designating legal liaisons (Wagner
et al. 2010), and creating a neutral setting to discuss sensitive information and resolve conflicts of
interest (Frueh et al. 2006).
MSCs comprise important operational and organizational scaffolds for collaborative innovation.
A key example is the concept of Safe Harbor applied within the VXDS program, which was pivotal in
building trust between industry and the FDA. The program allowed the successful development of the
pharmacogenomics data submission guidance documents, in which each draft was followed by a
collaborative FDA-industry meeting (Frueh et al. 2006). This collaboration led the pilot project for the
BQP, which was made possible by collaboration between Novartis and the FDA through the cooperative
R&D agreement (CRADA) as well as the Critical Path Institute (CPI). The CPI was created for the explicit
purpose of fostering precompetitive collaborations and data sharing and comprises a key example of the
subsequent shift towards precompetitive work (Clinton & Wechsler 2006; Woodcock 2010).
This shift was evident even internally, where the different divisions of the FDA-CDER, CDRH,
and CBER-collaborated as part of the IPRG review group (Gutman & Kessler 2006; Goodsaid & Frueh
2007). This organizational collaboration led to the development of the BQP, but more broadly, the FDA
divisions are more open to sharing knowledge and expertise. Although the BQP has developed mainly
Tingyu Liul40
under CDER, CDER is sharing knowledge that they have gained from implementing the BQP and working
with the various MSCs involved. Thus, CDRH learned from CDER's DDT and designed a similar initiative
termed the Medical Devices Development Tool (MDDT) (FDA 2013a).
As MSCs continue to evolve in breadth and scope, an important issue to be addressed is the
concept of global regulatory harmonization. So far, the FDA and EMA have collaborated at certain points
during both the VXDS and BQP process, with the future possibility to submit a joint biomarker
qualification package (Rinaldi 2011; Goodsaid & Mattes 2013). However, the two biomarker qualification
processes are not identical; for example, EMA requires a fee for submission and FDA does not, and the
latter has no formal deadlines for response times (Walker et al. 2013). In addition, the FDA and the EMA
have widely different regulations on IVDs; the FDA is generally considered more stringent (Senderowicz
& Pfaff 2014). Global harmonization in the case of the BQP process will be important to attain for
aligning on a topic of international consequence.
Finally, the increasing shift towards strategic partnerships across MSCs allows incremental value
generation from independent outputs and achievements. The evolution of the VXDS program to the
BQP, ADNI's growing Alzheimer's disease database, and PSTC's expansion beyond its preclinical focus by
collaborating with the Biomarkers Consortium and SAFE-T Consortium are all key examples (Dennis et al.
2013).
Section 5.4: Delivering tangible outputs
Many robust biomarkers and regulatory guidance documents that outline innovative policies for
biomarker adoption have already emerged from the work of MSCs. Qualified biomarkers include nine
nephrotoxicity biomarkers, one cardiac troponin biomarker, and three Alzheimer's disease-related
biomarkers. The EDRN has produced five FDA-approved diagnostics for early-stage cancer detection and
five diagnostics for use in CLIA-certified labs.
Moreover, a number of MSCs have contributed to the development of guidance documents that
are key in facilitating biomarker adoption. The VXDS program led to new policies on genomic data in the
form of regulatory guidance documents that both the FDA and industry found agreeable (Goodsaid et al.
2010). Furthermore, the MSCs involved in the BQP helped launch the DDT guidance document and
submitted biomarker packages that have been qualified for use in drug R&D. The Biomarkers
Consortium's I-SPY 2 trial contributed to the pCR draft guidance document, the "master IND" procedure,
and a regulatory pathway for CDx in the case of high-risk early breast cancer (L. J. Esserman et al. 2012).
However, although developing policies in a multi-stakeholder space can lend itself to increased
agreement and understanding, this is not always the case. For example, as of this writing, the IVD
guidance document is still open and controversial, with no agreement between the FDA and industry
(Carver 2010).
Section 5.5: Recommendation-Avoid fragmentation and duplication of preexisting efforts
One of the key drivers of future success for both collaborative and individual efforts is to avoid
fragmentation and duplication of work. An example of possible fragmentation is in the goals of the PSTC
and the Biomarkers Consortium. PSTC is clearly working towards biomarker qualification, but the
Biomarkers Consortium's work is less explicit. The Biomarkers Consortium has only stated that it plans to
take one of its biomarker projects--FDG-PET and volumetric CT for outcome measures for lung cancer
and lymphoma-through the qualification process at the FDA." Thus, although the PSTC and the
Biomarkers Consortium are collaborating to move qualified preclinical biomarkers into the clinical space
1 The final objectives of active projects at the Biomarkers Consortium can be found at:
http://www.biomarkersconsortium.org/projectsactive.php
Tingyu Liu 141
(Dennis et al. 2013), it is not clear if they are working towards the same goal of qualification. This can be
a point of confusion for other MSCs or individual stakeholders in the biomarker field. To give another
example, the EDRN has developed a rigorous data repository for biomarkers of early detection
(Srivastava 2013). However, it is unclear if the EDRN's work is being imparted to other MSCs working
towards similar rigorous standards. It is imperative for other MSCs who are building similar databases,
even if expanding beyond early detection biomarkers, to leverage the EDRN's learnings in terms of data
rigor and formatting.
It will be crucial for MSCs to be aware of other collaborative or individual efforts in the same
field and build upon each other's expertise. Even within the collaborative space, it is common to see
competition or conflicts of interests among MSCs. To avoid current fragmentation and duplication of
efforts, it will be important to have open, direct dialogue between the key members of each MSC. Such
conversations will serve to highlight not just the ultimate goals of their initiatives but also the necessary
consolidation of activities to achieve ultimate impact from the adoption of their outputs.
Section 5.6: Recommendation-Maintain flexible scientific processes
In the current environment of rapid scientific progress, scientific processes should be flexible, with the
ability to expand their scope as more data and information become available. While the current use of a
biomarker might be limited, it is good practice to keep the broader uses of the biomarker in mind and
have necessary infrastructure in place to capture relevant emerging data and knowledge. Such
information can be used to move the biomarker across subsequent stages of maturity (analytical validity
to clinical validity, for example) or across disease stages.
On this note, progressive knowledge expansion requires detailed documentation of the scientific
process to allow future efforts to build upon previous work. The PSTC and EDRN, for example, publish
their failures as well as successes in order to be comprehensive (Dennis et al. 2013). Efforts to improve
the accuracy of biomarker tests and integrate them into elaborate algorithms will be pivotal in achieving
consolidated information for drug treatment or preventive approaches.
Even further, the very process by which guidance documents and other such policies are created
and reviewed should be flexible as well. Dialogue needs to be kept open to make sure that policies
evolve with increased consensus of the affected stakeholders. As mentioned in Section 3, experts have
voiced concerns that the BQP team at the FDA has moved towards a more narrow context of use and a
conservative interpretation of the data, making the process more time-consuming and resourceintensive (Dennis et al. 2013). On an operational level, such concerns should be addressed in open
meetings between FDA and industry. To best ensure that such meetings occur and that policies continue
to evolve, MSCs should set firm rules for "check-in" meetings after certain milestones.
A key example of a progressive build-up from an existing guidance document is CAM D's novel
data-driven trial simulation model as a suitable clinical endpoint for describing cognitive changes in
patients with mild and moderate AD. The model underwent the same regulatory process as any other
biomarker and obtained a positive qualification decision from the EMA and the FDA. However, although
this model was not a "biomarker," FDA expanded the current definition of a DDT, rendering CAMD's
submission a "fit-for-purpose" drug development tool (Rogers 2013).
The DDT guidance document can accommodate further progressive expansion into other areas
of R&D as well as clinical practice. Experts have suggested that the DDT guidance can be adapted to
evaluate biomarkers for safety monitoring, personalized healthcare strategies, and post-market safety
surveillance (Dennis et al. 2013). As biomarker qualification expands in scope, further guidance will be
needed for designing acceptable retrospective studies (Khleif et al. 2010) and outlining expected
evidentiary standards (Campion et al. 2013), including stakeholder-wide consensus on the criteria for
analytical validity, clinical validity, and clinical utility. These next steps can be challenging in terms of
both necessary incentives and resources, highlighting the key role of MSCs to achieve agreement in
Tingyu Liu 142
purpose and prioritization of goals across different stakeholder groups. CRADAs, the industry-FDA
cooperative R&D platform, can be a good model to use, as the one-on-one factor provides for more
focused commitment.
In this context, it will be important, especially for regulators, to clarify the breadth and scope of
new policy paradigms such as the BQP. For example, although this was not stated by the regulators,
perhaps pCR as a surrogate endpoint did not undergo the BQP process because it is already described in
a draft guidance (FDA 2012) and is thus perceived as an accepted biomarker. If true, this would signify
that draft guidance documents can represent another biomarker adoption pathway. It would be helpful
for the learning process if the regulators explicitly stated that this was the case.
Section 5.7: Recommendation-Strategic and timely inclusion of more
stakeholders
The inclusion of more stakeholders, such as patients, diagnostic companies, and payers, diversifies the
decision criteria for an MSC. For example, patients and public advocacy groups can represent a powerful
force to push for biomarker development, by both bringing awareness and funding. However, without a
formal patient inclusion process in place, they still represent an "idiosyncratic process that depends on
the emergence of leadership in a given disease area" (Mittleman et al. 2013). Furthermore, given the
central role of patients in successful biomarker development and use, earlier and more encompassing
engagement should accelerate the field.
Diagnostic companies, on the other hand, have deep technical expertise on how to build the
analytically robust measurement devices that are necessary for the accurate performance of a
biomarker. However, diagnostics companies do not possess the commercial power of drug developers,
and what drug developers consider "precompetitive data" is typically considered "competitive" by
diagnostic companies. Perhaps for this reason, it has been difficult thus far to bring diagnostics
companies into biomarker collaborations, such as the I-SPY 2 trial, where there is little information on
the role they could play. It is possible that the drug developers involved in the trials have in-house
diagnostic divisions or have already acquired or partnered with diagnostic companies. However, such
approaches are not sustainable or sufficient to resolve the critical issues that are plaguing diagnostics
companies today in the commercial and regulatory space. Given the future impact of diagnostics in
personalized medicine, increased dialogue and awareness for every stakeholder group could help
resolve these challenges.
The engagement of payers is also of paramount importance because reimbursement decisions
play a major role in the actual use of biomarkers in the clinical setting. Although payers are starting to
demand clinical utility, evidence requirements are still vague (Quinn 2010). It is unclear how payers will
evaluate the biomarkers used in I-SPY 2 or the other biomarker projects from the Biomarkers
Consortium. In this respect, the Cancer Steering Committee Clinical Utility Working Group's efforts to
provide practical definitions for COU and resolve evidence requirements for IVDs will be critical
(Parkinson et al. 2014). The pilot process for moving an IVD jointly through the FDA and Center for
Medicare & Medicaid Services (CMS) represents an emerging example of payer engagement, but its
effectiveness remains to be seen, given its limited use (Hollmer 2013).
Although regulators represent a key stakeholder group in many of the MSCs we discussed, it will
be more beneficial to involve them earlier in discovery and development studies. Regulators can help
plan for the type of data required, at the same time gaining a deeper understanding of new technologies
and methodologies to better foster evidence-based regulatory decisions. Biomarker development is
moving into an era where the pace of innovation needs a new and more flexible regulatory mindset on
12
The Cancer Steering Committee Clinical Utility Working Group is a part of the Biomarkers
Consortium.
Tingyu Liu 143
how to evaluate the development process-from rigid decision points to progressive evaluations.
Regulators may have an interest in keeping decisions flexible so that they remain more in control of the
process.
Much like the earlier engagement of regulators, a later engagement of industry would be
desirable under certain situations, such as in the development of early detection biomarkers when the
commercial value may not be immediately apparent. At the EDRN, this approach is viewed as its core
strength; they have stated that, "Without the pressure of venture capital, 'best science' rather than
'best business' practice permits development of novel concepts that otherwise, because of commercial
needs, might never reach fruition" (Wagner & Srivastava 2012). The EDRN is funded mainly by NCI, an
organization that is anchored around strong scientific priorities. Although industry would be more
interested in the commercial outputs of scientific research, organizations like the EDRN can broaden the
focus from the end product to include the generation of knowledge on the complexity of disease and
physiology. Such new financial models can balance commercial incentives by supporting and
incentivizing more science-focused work.
Section 5.8: Recommendation-Develop metrics to track progress and
measure impact
Establishing metrics is important to set goals, manage collaborations, identify gaps and drive
improvements in any system. Innovation systems and management research have extensively discussed
how to measure the performance of a system and how the measurements must adapt to its lifecycle
(Carlsson et al. 2002). Especially in the field of collaborations, measurements must also be transparent
and subject to discussion and revision, so as to include all stakeholder viewpoints and avoid biased
behavior. For example, a metric such as "number of qualified biomarkers" may result in a large number
of biomarkers qualified based on a COU that is too narrow to be useful for sponsors; it will be necessary
for the MSCs involved to insure that qualified biomarkers provide value to drug developers.
Despite the extensive work with biomarker-focused MSCs, there is currently a lack of metrics
and performance measurement approaches. Some experts have criticized the Biomarkers Consortium
for its lack of output, remarking: "after 6 years in operation, does the Biomarkers Consortium represent
a good return on investment for its members? No qualified biomarker developed through the
Biomarkers Consortium has yet delivered tangible results to patients or other stakeholders" (Mittleman
et al. 2013). Developing metrics to track the progress and measure the impact of the Biomarkers
Consortium's deliverables will help them address such criticisms. Metrics are also broadly relevant to all
other ongoing collaborations and individual efforts to enable wider understanding in the biomarker
field.
A similar challenge is faced by the BQP-a key issue being the lack of transparency in the
program. For example, experts have stated that there is no public information about cycle time, the
process of how the BQP team assesses biomarkers, or its efficiency (Goodsaid & Mattes 2013). However,
certain members involved in the BQP are aware of the importance of such evaluations and have
suggested metrics that include "faster decision making for advancing or terminating developmental
compounds, increased numbers of compounds progressing through the pipeline, and reduced cost of
nonclinical and clinical development programs" (Walker et al. 2013). Moreover, the ability to track a
qualified biomarker's uptake is hindered by confidentiality provisions since biomarkers used in a drug
development program are considered to have moved back into the competitive space (Walker et al.
2013). An imperative for the sustainability of the qualification process is sharing information about
successes in biomarker uptake. Some members involved in the BQP feel that it is still too early to
develop metrics, especially to compare the BQP across agencies for strengths and weaknesses (Campion
et al. 2013). More data need to be gathered before informative analysis can occur. Such considerations
Tingyu Liu 144
led the EMA to agree that the BQP should be reviewed only after ten qualifications, when further
revision will be considered to adjust to the needs of all participants (Vaidya & Bonventre 2010).
Transparency is also an issue with metrics development for the EDRN. The EDRN's Data
Management and Coordination Center tracks a set of metrics which includes the number of
publications, patents, public-private partnerships, LDTs, and FDA-approved products (Crichton et al.
2010). However, industry's commercial rights prevent the EDRN from being completely transparent
about other crucial metrics, such as how EDRN-licensed products are performing in the marketplace.
Section 5.9: Conclusion
MSCs have introduced a variety of approaches that have the potential to significantly enhance our
capacity to effectively use biomarkers to optimize both drug R&D and clinical care. The analysis and
recommendations provided in this section serve to highlight the main gaps and opportunities, based on
selected key MSC examples. Tracking the development and progress of the various MSCs involved in
biomarker adoption has revealed many lessons for analysis and discussion. As MSCs are moving to the
next level, the different stakeholder groups should be careful to keep communication open to share
data, standards, scientific processes, and collaborative ideas, and to avoid fragmentation of efforts and
duplication of outcomes. It is imperative to keep in mind that all scientific and regulatory processes, new
or old, should be kept flexible, as technologies and science continue to evolve. As the magnitude of
challenges is too large for any single stakeholder, greater consensus and alignment will have to be
attained. On this front, it is important to involve stakeholders in a progressive and strategic manner with
MSCs, and to progressively engage new ones, including patients, diagnostic companies, and payers. Key
metrics will need to be developed to monitor performance and progress of individual and collaborative
efforts. The variety of stakeholder incentives and perceived conflicts of interests should be further
identified and formally considered.
During the past two decades, the biomedical industry has gradually shifted towards a more
collaborative approach to tackle challenges that exceed individual stakeholder capabilities and to
integrate the current fragmented paradigm in drug R&D. With careful design and implementation, this
crucial cultural shift will lead to greater openness and broader understanding to accelerate and drive
sustainable progress in biomedical innovation through the establishment of the necessary operational,
organizational, and management structures.
Tingyu Liu 145
Glossary A: The main biomarker categorizations proposed by field
experts
FDA
In order to try and establish a process for the submission and review of pharmacogenomics and
toxicogenomic data, the FDA developed a Guidance for Pharmacogenomics Data Submissions (2005)
(Goodsaid & Frueh 2006). The guidance defines biomarkers by its appropriateness in regulatory decision
making:
Probably valid
biomarker
A biomarker that is measured in an analytical test system with well-established
performance characteristics and for which there is a scientific framework or body
of evidence that appears to elucidate the physiologic, toxicologic, pharmacologic,
or clinical significance of the test results. A probable valid biomarker may not have
reached the status of a known valid marker because, for example, of any one of
the following reasons:
* The data elucidating its significance may have been generated within a
single company and may not be available for public scientific scrutiny.
* The data elucidating its significance, although highly suggestive, may not
be conclusive.
* Independent verification of the results may not have occurred.
Known valid
biomarker
A biomarker that is measured in an analytical test system with well-established
performance characteristics and for which there is widespread agreement in the
medical or scientific community about the physiologic, toxicologic, pharmacologic,
or clinical significance of the results.
The FDA, in its Guidance: Qualification Process for Drug Development Tools (2014) has defined
biomarkers by what disease and treatment characteristics they help define:
Diagnostic
A disease characteristic that categorizes a person by the presence or absence of a
specific physiological or pathophysiological state or disease.
Prognostic
A baseline characteristic that categorizes patients by degree of risk for disease
occurrence or progression of a specific aspect of a disease. A prognostic
biomarker informs about the natural history of the disorder in that particular
patient in the absence of a therapeutic intervention. It can be used as an
enrichment strategy to select patients likely to have clinical events of interest or
to progress rapidly.
Predictive
A baseline characteristic that categorizes patients by their likelihood of response
to a particular treatment relative to no treatment. A predictive biomarker can be
used as an enrichment strategy to identify a subpopulation likely to respond to a
treatment intervention in a particular way. It may predict a favorable response or
an unfavorable response (i.e., adverse event).
Pharmacodynamics
A change in the biomarker shows that a biological response has occurred in a
patient who has received a therapeutic intervention and for which the magnitude
of the change is considered pertinent to the response. A pharmacodynamic
biomarker may be treatment-specific or more broadly informative of disease
response.
Tingyu Liu 146
Industry
Experts from industry have defined biomarkers by graded categories based on the biomarker's intended
application (Wagner et al. 2007):
Demonstration
Biomarkers are associated with adequate preclinical sensitivity and specificity and
linked with clinical outcomes, but have not been reproducibly demonstrated in
Characterization
1"Surrogacy
clinical studies. Used for decision-making and supporting evidence with primary
clinical evidence in drug development. Corresponds to "Probable Valid Biomarker"
in FDA Guidance for Pharmacogenomics Data Submissions.
Biomarkers that have linked preclinical sensitivity and specificity with clinical
outcomes, and have been reproducibly demonstrated in more than 1 prospective
clinical study in humans. Used for decision-making, dose finding,
secondary/tertiary claims. Corresponds to "Known Valid Biomarker" in FDA
Guidance for Pharrnacogenomics Data Submissions.
Biomarkers that have undergone a holistic evaluation of the available data and can
substitute for a clinical endpoint. Used for registration in drug development. The
designation of "surrogate end point" requires agreement with regulatory
authorities.
_
__
Thomson Reuters
Many companies use the Integrity biomarker database from Thomson Reuters in their R&D process. This
database defines biomarkers more specifically, in order of continuum of care:
Risk stratification
Risk factor
The role of this biomarker is to determine a person's risk of suffering a particular
clinical event within a specified period of time.
The role of this biomarker is to determine a person's risk of a disease on the basis
of epidemiological evidence.
Disease profiling
The role of this biomarker is to obtain information about the disease, but there is
insufficient data to assign a clinical role. The data is often obtained from high
throughput analyses, for example transcriptional profiling, and might be
extrapolated to the processes that cause the disease.
Screening
The role of this biomarker is to sort a population into 'healthy' and 'non-healthy.'
Screening is an epidemiological process, though the same process may serve for
Diagnosis
Differential
diagnosis
Staging
The role of this biomarker is to identify or detect a disease
The role of this biomarker is to distinguish between two or more diseases with
similar signs and symptoms
The role of this biomarker is to describe how far a disease has progressed in a
patient. The stage at diagnosis is often a prognostic indicator of overall survival and
can be used as a guide for subsequent therapy.
Selection for
therapy
Toxicity profiling
The role of this biomarker is to select a sub-group of patients suitable for a
particular therapy.
The role of this biomarker is to obtain information about the underlying cause of
diagnosis as well.
Tingyu Liu 147
Prediction of drug
I resistance
Monitoring ot
treatment efficacy
The role of this biomarker is to detect possible resistance to therapy and thus to
exclude that theranv from the Dossible theranies available to the natient.
I
The role ot this biomarker is to identify signs of a change (usually beneficial) as a
result of treatment. A biomarker used for monitoring treatment efficacy is usually
measured before the treatment starts (baseline) in and at stages throughout the
treatment (follow up).
The Biomarkers Definitions Working Group
The Biomarkers Definitions Working Group (FDA, NIH, academia, industry, 2001) also provided a
classification system for biomarkers:
Type 0
(natural history)
Type I
(therapeutic
intervention)
Markers of natural history of a disease and that coordinate longitudinally with
known clinical indices, such as symptoms over the full range of disease states.
Markers that capture the effects of an intervention in accordance with the
mechanism of action of the drug, even though the mechanism might not be known
to be associated with clinical outcome.
]
Tingyu Liu 148
Glossary B: Mapping differences in semantics on the terms validation
and qualification
The terms validation and qualification are not well-defined. It has been stated: "The biomarker literature
occasionally uses 'validation' and 'qualification' or 'evaluation' interchangeably. We have avoided this
because the validation and qualification processes must be distinguished, and the term 'validation' does
not adequately describe the qualification process" (Wagner et al. 2007). The following tables highlight
key subject experts' views:
Lw11- onoGy vvm"5vvc "cpvuu~tiutc1 ailu attutG'
~(Bleavins 2010)
uOtteO
"biomarker validation [is] a continuum of technical bioanalytical method
validation, which is a familiar process that does not generally pose a problem
for an organization embarked on assay development"
Biornarkers
Definitions
Working Group
(Atkinson et al.
2001)
"the term validity is a broad concept that has been used to describe everything
from the analytical methods to the characteristics of the biomarkers identified.
Validity is also used across multiple industries, not only medical or health
disciplines. Therefore, when referring to biomarkers, validation is sometimes
termed qualification for clarity."
"The process of retrospectively linking a surrogate endpoint to a clinical
endpoint has often been referred to as validation. In addition, the term
validation is also often used to address performance characteristics (ie,
sensitivity, specificity, and reproducibility) of a measurement or an assay
technique... the term validation is unsuitable for the description of the process
(Lee et al. 2006)
of linking biomarkers to clinical endpoints."
"Fit-for-purpose method validation provides for efficient drug development by
conserving resources in the exploratory stages of biomarker characterization.
For example, a biomarker under exploratory development in an early phase
clinical trial would be less rigorously validated than an already well-qualified
biomarker in the same trial. By definition, exploratory biomarker data would
be used for less critical decisions"
(IOM 2010a)
(Lee et al. 2006)
Qualification can also be defined as a scientific linkage to disease: "A
description of the evidence relating to the biomarker in question-as
measured using validated tests-to the intervention and disease outcome"
Some have more narrowly refined qualification as clinical qualification: "The
evidentiary and statistical process linking biologic, pathologic and clinical
endpoints to the drug effect, or linking a biomarker to biologic and/or clinical
Tingyu Liu 149
endpoints.
(FDA 2014)
For the FDA, qualification aligns with regulatory decision and review through
the BQP: "A conclusion that within the stated context of use, the results of
assessment with a biomarker can be relied upon to have a specific
interpretation and application in drug development and regulatory review"
[BQP]
Tingyu Liu 150
Appendix A: Background information on the MSCs that have participated
in the BQP
Health and Environmental Science Institute (HESI) was established in 1989 as a global branch of the ILSI
(International Life Sciences Institute), with members from industry, academia, and regulatory agencies.
The committee's aims are to pool resources to support scientific research, sponsor symposia or
workshops, and conduct other technical activities. In 2002, it established a new program for biomarker
activity, with a focus on nephrotoxicity biomarkers. HESI was working towards peer-reviewed
publication, but when CPI met with them in March 2006 to discuss FDA's aim to establish a new process
for biomarker qualification review, HESI decided to submit data to FDA instead (Goodsaid & Mattes
2013). Although HESI played an instrumental role in the qualification of nephrotoxicity biomarkers and
cardiac troponin biomarkers, they have since decided that they will concentrate on biomarker discovery
instead and peer-reviewed publication as an endpoint (Goodsaid & Mattes 2013).
Predictive Safety Testing Consortium (PSTC) was formed by CPI in 2006, with members from industry,
EMA, FDA, and participating scientists, including clinicians and preclinical scientists. The Consortium's
mission is to collect and summarize data and assess drug safety with preclinical and clinical biomarkers.
PSTC is not oriented towards new biomarker discovery, but instead focuses on translational biomarkers
that the participants have at least some experience and confidence with (Sistare et al. 2010). PSTC
developed a "Progressive Qualification Framework" that defines the critical core set of data needed
initially, but also keeps dialogue and evidence gathering open, to allow regulatory agencies to anticipate
that additional data and further evaluations are expected and wanted (Sistare et al. 2010). The
Nephrotoxicity Working Group submitted full biomarker qualification package for seven biomarkers of
nephrotoxicity to the pilot process of biomarker qualification at FDA and EMA in July 2007 (Goodsaid et
al. 2008). These biomarkers were qualified in 2008 for non-clinical application in drug safety evaluation
using the rat model (Goodsaid & Mendrick 2010). These qualified biomarkers are now being tested for
use in safety tests in human clinical trials by the PSTC's Kidney Working Group and Translational Team,
collaborating with the Biomarker Consortium (Rinaldi 2011) as well as the Safer and Faster Evidencebased Translation Consortium (SAFE-T).
Coalition Against Major Diseases (CAMD) was launched in February 2008 with a mission to create new
tools and methods that can be applied during the development process of new treatments for
Alzheimer's and Parkinson's Disease. CAMD is focused on creating common data sharing standards,
establishing a database of pharmaceutical clinical trial data, developing disease progression models to
aid in clinical trial designs, and identifying biomarkers for use in patient selection (CAMD 2010). They are
basing their plan of action on PSTC's successful collaborative framework (Romero et al. 2009). CAMD
collaborated with Clinical Data Interchange Standards Consortium (CDISC) to publish the first data
standard for AD by pooling together data on more than 6000 patients who had been included in 21
clinical trials and combined this data with ADNI's data (Goldman et al. 2013). This is the first database of
combined clinical trials to be openly shared by companies to qualified researchers worldwide. Due in
part to their large-scale efforts, the FDA has mandated that CDISC's clinical standards will be required
for submissions by 2017 (Stephenson 2013). In addition, CAMD's work resulted in FDA issuing draft
guidance for enrichment strategies for clinical trials and developing drugs for treatment of early stage
AD (Stephenson 2013). CAMD has also submitted an AD simulation tool that was qualified by EMA and
FDA and a hippocampal volume measurement biomarker for clinical trial selection (qualified by the
EMA) (Sinha 2013). Currently, they have submitted a qualification package for the use of DAT imaging as
a biomarker to enrich clinical trials in patients with early-onset PD (Stephenson et al. 2013).
Tingyu Liu 151
The Alzheimer's Disease Neuroimaging Initiative (ADNI) began in 10/2004 as a joint business venture
between IND (a boutique CRO) and Synarc (an imaging CRO). Then, a large neuroimaging AD conference
brought together these companies with other like-minded industry stakeholders and they jointly
formulated the idea of industry co-funding a large gold standard AD study with NIH, thus leading to the
concept of ADNI (Khachaturian 2010). This background story demonstrates the power of bringing
different stakeholders together to promote innovation. ADNI initially encompassed clinical sites in the
US and Canada to identify neuroimaging measures and biomarkers with patients who have MCI and AD
(Mueller et al. 2005). ADNI's methods standardization is being built up by other consortia like the
Alzheimer's Association (Carrillo et al. 2013) and has since led to an umbrella organization called World
Wide ADNI for the different countries involved in the initiative, including Europe, China, and South Korea
(Burton 2011). Hundreds of scientists have made tens of thousands of downloads from ADNI's large
database (Miller 2009). Eli Lilly used ADNI-style clinical trials for their 2 AD compounds (Miller 2009) and
ADNI's work has contributed to recent revisions of diagnostic criteria for AD and MCI (Aisen 2011).
Tingyu Liu 152
Appendix B: Expert biographies
To substantiate our research and lend evidence to our findings, we had informal discussions with key
experts in the biomarker fields. Their biographies and some of the key issues discussed are listed below.
Shashi Amur, Ph.D
Shashi Amur is currently the Biomarker Qualification Science Coordinator in the office of Translational
Sciences, CDER, FDA. She received her Ph.D. in biochemistry from Indian Institute of Science, India and
completed post-doctoral fellowships at Temple University and at UCLA. She then gained experience in
diagnostic and biotech sectors before joining FDA as a senior genomics reviewer in the Office of Clinical
Pharmacology (OCP) and reviewed genomics- and biomarker-relevant sections of regulatory
submissions. She has been an invited speaker at national and international conferences, and the author
of 37 scientific publications. Her current research interests include pharmacogenomics, HLA-associated
adverse events and biomarkers in Autoimmune Diseases and in Alzheimer's disease. She has served as
Chair of the Pharmacogenomics Science Interest Group and Chair of OCP Science Day Committee at
OCP, CDER, FDA and has organized seminars and workshops in CDER, FDA.
Jonathan Fleming, M.S.
Jonathan Fleming is the Managing Partner of Oxford Bioscience Partners, an international venture
capital firm specializing in life science technology based investments, with offices in Boston,
Massachusetts and Seoul, Korea. He is a General Partner of the Korea-Seoul Life Science Fund. Mr.
Fleming has been in the investment business for over twenty seven years, starting and financing growth
companies in the United States, Europe, Israel and Asia. Prior to joining OBP in 1996, he was a Founding
General Partner of MVP Ventures in Boston, MA. He began his investment career with TVM Techno
Venture Management in Munich, Germany. Mr. Fleming has also co-founded Medica Venture Partners,
a venture capital investment firm specializing in early stage healthcare and biotechnology companies in
Israel. Mr. Fleming is a director of several public companies including Xencor (NASDAQ: XNCR) and
several private companies. He is also a director of Leerink Swann, a Boston based investment bank
specializing in healthcare companies. In addition, Mr. Fleming is a Senior Lecturer at the MIT Sloan
School of Business and a Member of the Board of NEHI, a healthcare oriented think tank and policy
advocacy group. Mr. Fleming is on the Chancellor's Commercialization Advisory Board of the University
of Texas System. Mr. Fleming holds a Master's degree in Public Administration from Princeton University
and a Bachelor of Arts degree from the University of California, Berkeley.
Bill Helming
Bill Helming is a Managing Director of Arryve Consulting. Helming's career has concentrated on
accelerating product development to help organizations gain competitive advantage and realize their
full strategic intent. His consulting engagements have focused on achieving synergies in operational
strategy, supply chain strategy and implementation, acquisition integration, R&D, technology transfer
and manufacturing scale-up. Helming has also worked extensively in the public sector, implementing
operational strategy to achieve important policy initiatives, such as strategic management for
government-funded research and development portfolios devoted to infectious disease
countermeasures for public health and biodefense. Prior to joining ARRYVE, Helming was partner and
led the Life Sciences practice at PRTM Consultants, where he worked for 21 years before it became a
division of PwC, where he was managing director. He earned his Master of Business Administration from
the Robert H. Smith School of Business at the University of Maryland, Master of Science in Engineering
from the Massachusetts Institute of Technology, and Bachelor of Science in Engineering from Webb
Institute.
Tingyu Liu
153
Huimin Kong, Ph.D
Huimin Kong serves as the President and Chief Executive Officer of Biohelix Corporation. BioHelix's
mission is to improve the quality of healthcare through the development of simple molecular diagnostic
tests for the near patient setting.
Andrea Schievella, Ph.D
Andrea Schievella is a scientist with 26 years of experience in cell and molecular biology, including
significant experience in licensing and business development with biotechnology and pharmaceutical
companies across the US and Europe. Her current role is a Business Manager at Cancer Research
Technology, where she facilitates licensing and collaborative arrangements between the non-profit
cancer charity Cancer Research UK and biotech and pharma companies working in cancer.
Sudhir Srivastava, Ph.D, MPH
Dr. Sudhir Srivastava has been Chief of the Cancer Biomarkers Research Group since 2000. His efforts
focus on molecular biology of malignancies, early malignancies, risk assessment, and informatics,
providing leadership in the areas of molecular screening and early detection. He is one of the principal
authors of the Bethesda Guidelines for diagnosing Hereditary non-polyposis colorectal cancer.
He received several national/international honors and awards and is a member of a number of scientific
committees. In 1995, he was elected to the American Joint Committee on Cancer, which is responsible
for developing staging criteria for cancers for worldwide use, and serves on the Executive Committeehe was the first Asian American and non-MD to do so. Under his leadership, AJCC accepted the inclusion
of tumor markers in the staging guidelines for colorectal cancer. He received several NIH and NCI honors
and awards, has initiated and chaired state-of-the-science national and international-level workshops
and conferences, and was principal architect of the first Gordon Research Conference on New Frontiers
in Cancer Detection and Diagnosis. He initiated new areas of research (molecular signatures of infectious
agents in cancer; micro-imaging in classifying preneoplastic lesions; nano-technology in Earlier Cancer
Detection; Metabolomics and Glycomics Alliances with other NIH Institutes). He published more than
170 peer-reviewed papers and has edited four books. He is Editor-In-Chief for the journal Disease
Biomarkers; led creation of the journal Cancer Biomarkers; was appointed to the editorial board of the
journal Cancer Prevention Research; is a member of several scientific committees; founding member of
HUPO; principal architect of NCI's Early Detection Research Network; and founding Editor and Board
member for Journal of Proteomics & Bioinformatics. He has a PhD in biological science; MS in computer
science; and MPH from the Johns Hopkins University.
Mark Trusheim, M.S.
Mark Trusheim is founder of Co-Bio Consulting, LLC as well as Executive in Residence and Visiting Scholar
at the MIT Sloan School of Management and a Special Government Employee for the FDA's Office of the
Commissioner. He is a former member of the Massachusetts Biotechnology Council's Board of Directors,
which helps its over 500 members succeed in the state. In 2004 he further served as the Interim
President of the MBC, leading its successful legislative agenda, its expansion of MassBioEd education
programs and its continued membership growth. As an entrepreneur, Mark founded and was the first
President and CEO of Cantata Laboratories. Cantata marketed clinical diagnostics and pharmaceutical
biomarker services based on its biochemical profiling platform. Prior to Cantata, Mark spent over 10
years at Monsanto/Pharmacia, culminating his career there as Co-President and Chief Operating Officer
of Cereon Genomics, LLC. He holds degrees in Chemistry from Stanford University and Management
from MIT.
Tingyu Liu 154
Andrey Zarur
Andrey Zarur is managing partner at Kodiak Venture Partners (KVP), a venture capital firm specializing in
formation stage information and life technology investments. He has been active in early stage life
sciences companies for over 15 years, and has actively participated in the creation of over a dozen
companies in the healthcare and clean energy sectors. Prior to joining KVP, he was founder and chief
executive officer of BioProcessors, a leading provider of high-throughput systems to the
biopharmaceutical industry. In addition to BioProcessors, he has led four life science companies from
inception to exit, and has been an active investor or board member on several others. Zarur is a cofounder and chairman of the board of GreenLight Biosciences and Lumicell. He also is chairman of the
board of Allegro Diagnostics. Zarur is a director of several private companies, including Fluxion
Biosciences, Astadia, and WTI, and is a director of several nonprofit organizations such as Infantia, a
foundation dedicated to providing education to children in developing countries. Zarur holds a BS from
the National University of Mexico and an SM and a PhD in chemical engineering from MIT.
Indicative Questions
1) What do you think are the biggest gaps/bottlenecks in the introduction of biomarkers in R&D
and clinical practice?
2) What impact do you think the different stakeholders can have in terms of addressing these
challenges?
3) How do you think the different stakeholders can be brought to work closer together?
4) In which ways do you think MSCs are useful? In which ways could they be improved? What are
some metrics of success that you think would be important to evaluate these MSCs by?
5) Do you think the lack of harmonization in biomarker terminology and standards have hindered
the adoption of innovative biomarkers in drug development and patient management?
6) Biomarker development is a very data-rigorous field. What would be the best way to address
such challenges?
7) What are your thoughts on the BQP? Do you think that it can make a difference in providing a
better regulatory path for biomarker use?
8) What are your thoughts on the reason behind the regulatory differences between the FDA and
EMA?
Tingyu Liul55
Bibliography
Advameddx 2013, 'Infographic: Diagnostics Transforming Care'. Advameddx. Available at:
http://advameddx.org/go.cfm?do=News.View&nid=14 [Accessed April 23, 2014].
Aisen, PS 2011, 'Q&A: The Alzheimer's Disease Neuroimaging Initiative.'. BMC Medicine, 9, p.101.
Allison, M 2010, 'Biomarker-led adaptive trial blazes a trail in breast cancer.'. Nature Biotechnology,
28(5), pp.383-4.
Altar, C, Amakye, D, Bounos, D, Bloom, J, Clack, G, Dean, R, Devanarayan, V, Fu, D, Furlong, S, Hinman, L,
Girman, C, Lathia, C, Lesko, L, Madani, S, Mayne, J, Meyer, J, Raunig, D, Sager, P, Williams, S, Wong,
P & Zerba, K 2008, 'A prototypical process for creating evidentary standards for biomarkers and
diagnostics'. Clinical Pharmacology & Therapeutics, 83(2), pp.368-371.
Amur, S, Frueh, F, Lesko, L & Huang, S 2008, 'Integration and use of biomarkers in drug development,
regulation and clinical practice: a US regulatory perspective'. Biomarkers in Medicine, 2(3), pp.305311.
Anderson, D & Kodukula, K 2013, 'Biomarkers in pharmacology and drug discovery'. Biochemical
Pharmacology, 87(1), pp.172-88.
Anon 2013, 'Medical Device Innovation Consortium'. Available at: http://mdic.org/ [Accessed April 27,
2014].
Atkinson, A, Colburn, W, DeGruttola, V, DeMets, D, Downing, G, Hoth, D, Oates, J, Peck, C, Schooley, R,
Spilker, B, Woodcock, J & Zeger, S 2001, 'Biomarkers and surrogate endpoints: preferred
definitions and conceptual framework.'. Clinical Pharmacology & Therapeutics, 69(3), pp.89-95.
Baggerly, KA & Coombes, KR 2011, 'What information should be required to support clinical "omics"
publications?'. Clinical Chemistry, 57(5), pp.688-90.
Barker, A, Sigman, C, Kelloff, G, Hylton, N, Berry, D & Esserman, U 2009, 'I-SPY 2: an adaptive breast
cancer trial design in the setting of neoadjuvant chemotherapy.'. Clinical Pharmacology &
Therapeutics, 86(1), pp.97-100.
Barratt, RA, Bowens, SL, McCune, SK, Johannessen, JN & Buckman, SY 2012, 'The critical path initiative:
leveraging collaborations to enhance regulatory science.'. Clinical Pharmacology & Therapeutics,
91(3), pp.380-3.
Bleavins, MR 2010, 'Biomarkers in Drug Development', Wiley.
Burton, A 2011, 'Big science for a big problem: ADNI enters its second phase.'. Lancet Neurology, 10(3),
pp.206-7.
CAMD 2010, 'FACT SHEET: Coalition Against Major Diseases', Tuscon, AZ. Available at: http://cpath.org/pdf/CoalitionFactSheet.pdf.
Tingyu Liu 156
Campion, S, Aubrecht, J, Boekelheide, K, David, W, Vaidya, VS, Anderson, L, Burt, D, Hwang, K, Pacheco,
S & Saikumar, J 2013, 'The current status of biomarkers for predicting toxicity'. Expert Opinion on
Drug Metabolism & Toxicology, 9(11), pp.1-28.
Carlsson, B, Jacobsson, S, Holmen, M & Rickne, A 2002, 'Innovation systems: analytical and
methodological issues'. Research Policy, 31, pp.233-245.
Carrillo, M, Blennow, K, Soares, H, Lewczuk, P, Mattsson, N, Oberoi, P, Umek, R, Vandijck, M, Salamone,
S, Bittner, T, Shaw, L, Stephenson, D, Bain, L & Zetterberg, H 2013, 'Global standardization
measurement of cerebral spinal fluid for Alzheimer's disease: An update from the Alzheimer's
Association Global Biomarkers Consortium'. Alzheimer's & Dementia, 9, pp.137-140.
Carroll, J 2012, 'The Top Phase Ill Disasters of 2012'. FierceBiotech. Available at:
http://www.fiercebiotech.com/special-report/top-phase-iii-disasters-2012/2012-10-02 [Accessed
March 9, 2014].
Carver, K 2010, 'Companion Diagnostics: Evolving FDA Regulation and Issues for Resolution'. In S. Danzis
& E. Flannery, eds. In Vitro Diagnostics: The Complete Regulatory Guide. The Food and Drug Law
Institute, pp. 149-184.
CDRH 2010, 'FDA/CDRH Public Meeting: Oversight of Laboratory Developed Tests (LDTs)'. In Workshops
& Conferences (Medical Devices). Center for Devices and Radiological Health. Available at:
http://www.fda.gov/MedicalDevices/NewsEvents/WorkshopsConferences/ucm212830.htm
[Accessed March 27, 2014].
Chataway, J, Fry, C, Marjanovic, S & Yaqub, 0 2012, 'Public-private collaborations and partnerships in
stratified medicine: making sense of new interactions'. New Biotechnology, 29(6), pp.732-40.
Chen, M, Vijay, V, Shi, Q, Liu, Z, Fang, H & Tong, W 2011, 'FDA-approved drug labeling for the study of
drug-induced liver injury.'. Drug Discovery Today, 16(15-16), pp.697-703.
Clinton, BP & Wechsler, J 2006, 'What Ever Happened to Critical Path'. Pharmaceutical Executive, pp.17.
Crichton, DJ, Mattmann, CA, Thornquist, M, Anton, K & Hughes, JS 2010, 'Bioinformatics: biomarkers of
early detection.'. Cancer Biomarkers, 9(1-6), pp.511-30.
Day, M, Rutkowski, J & Feuerstein, G 2009, 'Translational Medicine--A Paradigm Shift in Modern Drug
Discovery and Development: The Role of Biomarkers'. In C. Guzman & G. Feuerstein, eds.
Pharmaceutical Biotechnology. pp. 1-12.
Dennis, EH, Walker, EG, Baker, AF & Miller, RT 2013, 'Opportunities and Challenges of Safety Biomarker
Qualification: Perspectives from the Predictive Safety Testing Consortium'. Drug Development
Research, 74(2), pp.112-126.
Dieterle, F, Sistare, F, Goodsaid, F, Papaluca, M, Ozer, J, Webb, C, Baer, W, Senagore, A, Schipper, M,
Vonderscher, J, Sultana, S, Gerhold, D, Phillips, J, Maurer, G, Carl, K, Laurie, D, Harpur, E, Sonee, M,
Tingyu Liu 157
Ennulat, D, Holder, D, Andrews-Cleavenger, D, Gu, Y, Thompson, K, Goering, P, Vidal, J, Abadie, E,
Maciulaitis, R, Jacobson-Kram, D, Defelice, A, Hausner, E, Blank, M, Thompson, A, Harlow, P,
Throckmorton, D, Xiao, S, Xu, N, Taylor, W, Vamvakas, S, Flamion, B, Lima, B, Kasper, P, Pasanen,
M, Prasad, K, Troth, S, Bounous, D, Robinson-Gravatt, D, Betton, G, Davis, M, Akunda, J, McDuffie,
J, Suter, L, Obert, L, Guffroy, M, Pinches, M, Jayadev, S, Blomme, E, Beushausen, S, Barlow, V,
Collins, N, Waring, J, Honor, D, Snook, S, Lee, J, Rossi, P, Walker, E & Mattes, W 2010, 'Renal
biomarker qualification submission: a dialog between the FDA-EMEA and Predictive Safety Testing
Consortium.'. Nature Biotechnology, 28(5), pp.455-62.
Drucker, E & Krapfenbauer, K 2013, 'Pitfalls and limitations in translation from biomarker discovery to
clinical utility in predictive and personalised medicine.'. The EPMA Journal, 4(1), p.7.
EGAPP 2014, 'The EGAPP initiative: lessons learned.'. Genetics in Medicine, 16(3), pp.217-24.
EMA 2006, 'Guideline on Pharmacogenetics Briefing Meetings', London, UK. Available at:
http://www.ema.europa.eu/docs/en_GB/document-library/Scientificguideline/2009/09/WC5000
03886.pdf.
EMA 2014, 'Pharmacogenomics Working Party'. EMA. Available at:
http://www.ema.europa.eu/ema/index.jsp?curl=pages/contacts/CH MP/peoplelisting_000018.jsp
&mid=WC0b01ac0580028d91 [Accessed April 7, 2014].
Esserman, L, Berry, D, DeMichele, A, Carey, L, Davis, S, Buxton, M, Hudis, C, Gray, JW, Perou, C, Yau, C,
Livasy, C, Krontiras, H, Montgomery, L, Tripathy, D, Lehman, C, Liu, MC, Olopade, 01, Rugo, HS,
Carpenter, JT, Dressier, L, Chhieng, D, Singh, B, Mies, C, Rabban, J, Chen, Y, Giri, D, van 't Veer, L &
Hylton, N 2012, 'Pathologic complete response predicts recurrence-free survival more effectively
by cancer subset: results from the I-SPY 1 TRIAL'. Journal of Clinical Oncology, 30(26), pp.3242-9.
Esserman, U, Barker, AD, Woodcock, J, Buxton, MA, Berry, DA, Patterson, R, Jolly, MK, Demichele, A,
Hylton, N, Rubin, EH, Parkinson, D, Wholley, D, van 't Veer, L, Yee, D, Park, J, Tripathy, D,
Perlmutter, J, Buetow, K, Hogarth, M, Gray, J & Dilts, DM 2012, 'A Model for Accerlerating
Identification and Regulatory Approval of Effective Investigational Agents'. Cureus, 4(12), pp.1-17.
Esserman, U, Buxton, M & Berry, DA 2013, 'Accelerating approval of promising agents in the
neoadjuvant setting: a path for facilitating targeted drug development for breast cancer'. Breast
Diseases: A Year Book Quarterly, 24(2), pp.126-128.
Esserman, U & Woodcock, J 2011, 'Accelerating Identification and Regulatory Approval of Investigational
Cancer Drugs'. JAMA: The Journal of the American Medical Association, 306(23), pp.2608-2609.
Etzkowitz, H & Leydesdorff, L 1995, 'The Triple Helix: University-Industry-Government Relations: A
Laboratory for Knowledge-Based Economic Development'. EASST Review, 14, pp.14-19.
Evans, BJ 2010, 'Establishing clinical utility of pharmacogenetic tests in the post-FDAAA era.'. Clinical
Pharmacology & Therapeutics, 88(6), pp.749-51.
Tingyu Liu 158
FDA 2014, 'Guidance for Industry and FDA Staff: Qualification Process for Drug Development Tools',
Silver Spring, MD. Available at:
http://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatorylnformation/Guidances/UC
M230597.pdf.
FDA 2012, 'Guidance for Industry: Pathologic Complete Response in Early-Stage Breast Cancer: Use as an
Endpoint to Support Accelerated Approval', Available at:
http://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatorylnformation/Guidances/UC
M305501.pdf.
FDA 2013a, 'Medical Device Development Tools: Draft Guidance for Industry, Tool Developers, and Food
and Drug Administration Staff, Available at:
http://www.fda.gov/downloads/MedicalDevices/DeviceRegulationandGuidance/GuidanceDocume
nts/UCM374432.pdf.
FDA 2010, 'Number of voluntary exploratory data submissions (VXDS) and biomarker qualification
reviews completed per month'. FDA-TRACK: Agency-wide Program Performance. Available at:
http://www.accessdata.fda.gov/FDATrack/track?program=cder&id=CDER-OCP-VXDS-ReviewsCompleted [Accessed April 7, 2014].
FDA 2013b,'Paving the Way for Personalized Medicine', Available at:
http://www.fda.gov/downloads/scienceresearch/specialtopics/personalizedmedicine/ucm372421.
pdf.
Fox, JL 2014, 'Master Protocol for squamous cell lung cancer readies for launch.'. Nature Biotechnology,
32(2), pp.116-8.
Frueh, FW, Rudman, A, Simon, K, Gutman, S, Reed, C & Dorner, AJ 2006, 'Experience with voluntary and
required genomic data submissions to the FDA: summary report from track 1 of the third FDA-DIAPWG-PhRMA-BIO pharmacogenomics workshop.'. The Pharmacogenomics Journal, 6(5), pp.296-
300.
Gerard, B, Betard, M, Smith, B & Denyer, M 2012, 'Development of targeted agents and companion
diagnostics'. Drug Information Journal, 46(4), pp.405-419.
Gibbs, J 2010, 'Regulatory Pathways for Clearance or Approval of IVDs'. In S. Danzis & E. Flannery, eds. In
Vitro Diagnostics: The Complete Regulatory Guide. Food and Drug Law Institute, pp. 43-68.
Goldman, M 2012, 'The innovative medicines initiative: a European response to the innovation
challenge.'. Clinical Pharmacology & Therapeutics, 91(3), pp.418-25.
Goldman, M, Compton, C & Mittleman, BB 2013, 'Public-private partnerships as driving forces in the
quest for innovative medicines'. Clinical and Translational Medicine, 2(2), pp.1-3.
Goodsaid, F, Amur, S, Aubrecht, J, Burczynski, M, Carl, K, Catalano, J, Charlab, R, Close, S, Cornu-Artis, C,
Essioux, L, Fornace, A, Hinman, L, Hong, H, Hunt, I, Jacobson-Kram, D, Jawaid, A, Laurie, D, Lesko, L,
Li, H, Lindpaintner, K, Mayne, J, Morrow, P, Papaluca-Amati, M, Robison, T, Roth, J, Schuppe-
Tingyu Liu
159
Koistinen, I, Shi, L, Spleiss, 0, Tong, W, Truter, S, Vonderscher, J, Westelinck, A, Zhang, L & Zineh, I
2010, 'Voluntary exploratory data submissions to the US FDA and the EMA: experience and
impact.'. Nature Reviews. Drug Discovery, 9(6), pp.435-45.
Goodsaid, F & Frueh, F 2007, 'Biomarker Qualification Pilot Process at the US Food and Drug
Administration'. AAPS Journal, 9(1), pp.105-108.
Goodsaid, F & Frueh, F 2006, 'Process map proposal for the validation of genomic biomarkers'.
Pharmacogenomics, 7(5), pp.773-782.
Goodsaid, F, Frueh, F & Mattes, W 2008, 'Strategic paths for biomarker qualification'. Toxicology, 245,
pp.219-223.
Goodsaid, F & Mattes, W 2013,'The Path from Biomarker Discovery to Regulatory Qualification', Elsevier
Inc.
Goodsaid, F & Mendrick, D 2010, 'Translational medicine and the value of biomarker qualification.'.
Science Translational Medicine, 2(47), pp.1-3.
Goodsaid, F & Papaluca, M 2010, 'Evolution of biomarker qualification at the health authorities.'. Nature
Biotechnology, 28(5), pp.441-3.
Gutman, S & Kessler, L 2006, 'The US Food and Drug Administration perspective on cancer biomarker
development.'. Nature Reviews. Cancer, 6(7), pp.565-71.
Halim, A 2011, 'Biomarkers in Drug Development: A Useful Tool but Discrepant Results May Have a
Major Impact'. In Drug Discovery and Development - Present and Future. pp. 401-424. Available at:
http://www.intechopen.com/books/drug-discovery-and-development-present-andfuture/biomarkers-in-drug-development-a-useful-tool-but-discrepant-results-may-have-a-majorimpact.
Halim, A 2009, 'Impact of discrepant results from clinical laboratories on patients and pharmaceutical
trials: evidence from proficiency testing results'. Biomarkers in Medicine, 3(3), pp.231-238.
Hausner, EA, Hicks, KA, Leighton, JK, Szarfman, A, Thompson, AM & Harlow, P 2013, 'Qualification of
cardiac troponins for nonclinical use: A regulatory perspective'. Regulatory Toxicology and
Pharmacology, 67, pp.108-114.
Hayes, DF, Allen, J, Compton, C, Gustavsen, G, Leonard, DGB, McCormack, R, Newcomer, L, Pothier, K,
Ransohoff, D, Schilsky, RL, Sigal, E, Taube, SE & Tunis, SR 2013, 'Breaking a vicious cycle.'. Science
Translational Medicine, 5(196), pp.1-7.
Hockett, RD & Close, SL 2010, 'Regulation of laboratory-developed tests: the case for utilizing
professional associations.'. Clinical Pharmacology & Therapeutics, 88(6), pp.743-5.
Hollmer, M 2013, 'FDA, CMS extend little-used pilot program that grants parallel review for some
devices'. FierceMedicalDevices. Available at: http://www.fiercemedicaldevices.com/story/fda-cms-
Tingyu Liu 160
extend-little-used-pilot-program-grants-parallel-review-some-device/2013-12-19 [Accessed March
27, 2014].
IOM 2010a,'Evaluation of Biomarkers and Surrogate Endpoints in Chronic Disease' C. M. Micheel & J. R.
Ball, eds., National Academies Press.
IOM 2010b, 'Extending the Spectrum of Precompetitive Collaboration in Oncology Research: Workshop
Summary' IOM, ed., National Academies Press.
IOM 2008, 'Neuroscience Biomarkers and Biosignatures', The National Academies Press.
Keeling, P & Roth, M 2008,'IP Considerations in Early Biomarker Development in the Personalized
Medicine Age', Available at: http://www.diaceutics.com/sites/default/files/publications/IP
considerations in early Biomarker development in PM.pdf.
Kelloff, GJ & Sigman, CC 2012, 'Cancer biomarkers: selecting the right drug for the right patient.'. Nature
Reviews. Drug Discovery, 11(3), pp.201-14.
Khachaturian, ZS 2010, 'Perspective on the Alzheimer's disease neuroimaging initiative: progress report
and future plans.'. Alzheimer's & Dementia, 6(3), pp.199-201.
Khleif, SN, Doroshow, JH & Hait, WN 2010, 'AACR-FDA-NCI Cancer Biomarkers Collaborative consensus
report: advancing the use of biomarkers in cancer drug development.'. Clinical Cancer Research,
16(13), pp.3299-318.
Lee, J, Devanarayan, V, Barrett, YC, Weiner, R, Allinson, J, Fountain, S, Keller, S, Weinryb, I, Green, M,
Duan, L, Rogers, J a, Millham, R, O'Brien, PJ, Sailstad, J, Khan, M, Ray, C & Wagner, J 2006, 'Fit-forpurpose method development and validation for successful biomarker measurement.'.
Pharmaceutical Research, 23(2), pp.312-28.
Lesko, L 2007, 'Paving the Critical Path: How can Clinical Pharmacology Help Achieve the Vision?'. Clinical
Pharmacology & Therapeutics, 81(2), pp.170-177.
Lesko, L & Atkinson, A 2001, 'Use of biomarkers and surrogate endpoints in drug development and
regulatory decision making: criteria, validation, strategies'. Annual Review of Pharmacology
Toxicology, 41, pp.347-66.
Mahajan, R & Gupta, K 2010, 'Food and drug administration's critical path initiative and innovations in
drug development paradigm: Challenges, progress, and controversies.'. Journal of Pharmacy &
Bioallied Sciences, 2(4), pp.307-13.
Manolis, E, Vamvakas, S & Isaac, M 2011, 'New pathway for qualification of novel methodologies in the
European medicines agency'. Proteomics. Clinical Application, 5, pp.248-255.
Mansfield, EA 2014, 'FDA Perspective on Companion Diagnostics: An Evolving Paradigm'. Clinical Cancer
Research, 20(6), pp.1453-1457.
Tingyu Liul61
Marrer, E & Dieterle, F 2007, 'Promises of Biomarkers in Drug Development - A Reality Check'. Chemical
Biology & Drug Design, 69, pp.381-394.
Mattes, WB, Walker, EG, Abadie, E, Sistare, FD, Vonderscher, J, Woodcock, J & Woosley, RL 2010,
'Research at the interface of industry, academia and regulatory science'. Nature Biotechnology,
28(5), pp.432-433.
Miller, G 2009, 'Alzheimer's biomarker initiative hits its stride'. Science, 326, pp.386-389.
Mittleman, B, Neil, G & Cutcher-Gershenfeld, J 2013, 'Precompetitive consortia in biomedicine--how are
we doing?'. Nature Biotechnology, 31(11), pp.979-85.
Mueller, SG, Weiner, MW, Thai, U, Petersen, RC, Jack, C, Jagust, W, Trojanowski, JQ, Toga, AW &
Beckett, L 2005, 'The Alzheimer's disease neuroimaging initiative.'. Neuroimaging Clinics of North
America, 15(4), pp.869-77, xi-xii.
Mullard, A 2014a, '2013 FDA drug approvals.'. Nature reviews. Drug discovery, 13(2), pp.85-9.
Mullard, A 2014b, 'Multicompany trials adapt to disciplines beyond cancer.'. Nature medicine, 20(1), p.3.
NCI 2012, 'NCI-FDA-NIST Workshop on Standards in Molecular Diagnostics for the Discovery and
Validation of Clinically Useful Cancer Biomarkers', Rockville, MD. Available at:
http://prevention.cancer.gov/files/groups/NCI-FDA-NISTBiomarkersWorkshopReport.pdf.
Opar, A 2011, 'Bridging the drug-diagnostic divide.'. Nature Reviews. Drug Discovery, 10(5), pp.323-4.
Orr, M, Goodsaid, F, Amur, S, Rudman, A & Frueh, F 2007, 'The experience with voluntary genomic data
submissions at the FDA and a vision for the future of the voluntary data submission program.'.
Clinical Pharmacology & Therapeutics, 81(2), pp.294-7.
Pammolli, F, Magazzini, L & Riccaboni, M 2011, 'The productivity crisis in pharmaceutical R&D.'. Nature
Reviews. Drug Discovery, 10(6), pp.428-38.
Parkinson, DR, McCormack, RT, Keating, SM, Gutman, SI, Hamilton, SR, Mansfield, EA, Piper, MA,
DeVerka, P, Frueh, FW, Jessup, JM, McShane, LM, Tunis, SR, Sigman, CC & Kelloff, GJ 2014,
'Evidence of Clinical Utility: An Unmet Need in Molecular Diagnostics for Patients with Cancer'.
Clinical Cancer Research, 20(6), pp.1428-1444.
Patlak, M 2010, 'Competitors try collaboration to speed drug development.'. Journal of the National
Cancer Institute, 102(12), pp.841-3.
Paul, SM, Mytelka, DS, Dunwiddie, CT, Persinger, CC, Munos, BH, Lindborg, SR & Schacht, AL 2010, 'How
to improve R&D productivity: the pharmaceutical industry's grand challenge.'. Nature Reviews.
Drug Discovery, 9(3), pp.203-14.
Pennello, G 2013, 'Analytical and clinical evaluation of biomarkers assays: when are biomarkers ready
for prime time?'. Clinical Trials, 10(5), pp.666-76.
Tingyu Liu 162
Pepe, MS, Etzioni, R, Feng, Z, Potter, JD, Lou, M, Thornquist, M, Winget, M & Yasui, Y 2001, 'Phases of
Biomarker Development for Early Detection of Cancer'. Journal of the National Cancer Institute,
93(14), pp.1054-1061.
Pignatti, F, Ehmann, F, Hemmings, R, Jonsson, B, Nuebling, M, Papaluca-Amati, M, Posch, M & Rasi, G
2014, 'Cancer Drug Development and the Evolving Regulatory Framework for Companion
Diagnostics in the European Union'. Clinical Cancer Research, 20(6), pp.1458-1468.
Poste, G 2012, 'Biospecimens, biomarkers, and burgeoning data : the imperative for more rigorous
research standards'. Trends in Molecular Medicine, 18(12), pp.718-723.
Poste, G, Carbone, DP, Parkinson, DR, Verweij, J, Hewitt, SM & Jessup, JM 2012, 'Leveling the playing
field: bringing development of biomarkers and molecular diagnostics up to the standards for drug
development.'. Clinical Cancer Research, 18(6), pp.1515-23.
Prensner, JR, Chinnaiyan, AM & Srivastava, S 2012, 'Systematic, evidence-based discovery of biomarkers
at the NCI.'. Clinical & Experimental Metastasis, 29(7), pp.645-52.
PRNewswire 2013, 'Vermillion's OVA1 Receives New Statement by Society of Gynecologic Oncology'.
PRNewswire. Available at: http://www.prnewswire.com/news-releases/verm illions-oval-receivesnew-statement-by-society-of-gynecologic-oncology-210984641.html [Accessed April 14, 2014].
Prowell, T & Pazdur, R 2012, 'Pathological complete response and accelerated drug approval in early
breast cancer'. New England Journal of Medicine, 366(26), pp.2438-2441.
Quinn, B 2010, 'Payers and the assessment of clinical utility for companion diagnostics.'. Clinical
Pharmacology & Therapeutics, 88(6), pp.751-4.
Ranga, M & Etzkowitz, H 2013, 'Triple Helix systems: an analytical framework for innovation policy and
practice in the Knowledge Society'. Industry and Higher Education, 27(4), pp.237-262.
Ray, T 2013, 'After Long Silence on LDT Regulation, FDA Commish Revives Thorny Topic at ASCO'.
Phamacogenomics Reporter. Available at: http://www.genomeweb.com/after-long-silence-idtregulation-fda-commish-revives-thorny-topic-asco.
Rinaldi, A 2011, 'Teaming up for biomarker future'. EMBO Reports, 12(6), pp.500-4.
Rogers, M 2013, 'AD Trial Simulation Tool Receives Regulators' Blessings'. Alzforum. Available at:
http://www.alzforum.org/news/research-news/ad-trial-simulation-tool-receives-regulatorsblessings?id=3542 [Accessed April 7, 2014].
Romero, K, Mars, MD, Frank, D, Anthony, M, Neville, J, Kirby, L, Smith, K & Woosley, RL 2009, 'The
Coalition Against Major Diseases: Developing Tools for an Integrated Drug Development Process
for Alzheimer's and Parkinson's Diseases'. Clinical Pharmacology & Therapeutics, 86(4), pp.365-
368.
Tingyu Liu 163
Schully, SD, Benedicto, CB, Gillanders, EM, Wang, SS & Khoury, MJ 2011, 'Translational research in
cancer genetics: the road less traveled.'. Public Health Genomics, 14(1), pp.1-8.
Senderowicz, AM & Pfaff, 0 2014, 'Similarities and Differences in the Oncology Drug Approval Process
between FDA and European Union with Emphasis on In Vitro Companion Diagnostics.'. Clinical
Cancer Research, 20(6), pp.1445-52.
Sinha, V 2013,'Model-based Drug Development in Neurodegenerative Diseases: Regulatory
Perspectives', Available at: http://c-path.org/wp-content/uploads/2013/1/Session-l_CAMDAnnual-Meeting_2013_Sinha.pdf.
Sistare, F, Dieterle, F, Troth, S, Holder, D, Gerhold, D, Andrews-Cleavenger, D, Baer, W, Betton, G,
Bounous, D, Carl, K, Collins, N, Goering, P, Goodsaid, F, Gu, Y, Guilpin, V, Harpur, E, Hassan, A,
Jacobson-Kram, D, Kasper, P, Laurie, D, Lima, B, Maciulaitis, R, Mattes, W, Maurer, G, Obert, L,
Ozer, J, Papaluca-Amati, M, Phillips, J, Pinches, M, Schipper, M, Thompson, K, Vamvakas, S, Vidal, J,
Vonderscher, J, Walker, E, Webb, C & Yu, Y 2010, 'Towards consensus practices to qualify safety
biomarkers for use in early drug development'. Nature Biotechnology, 28(5), pp.446-455.
Srivastava, S 2013, 'The early detection research network: 10-year outlook.'. Clinical Chemistry, 59(1),
pp.60-7.
Stephenson, D 2013, 'Coalition Against Major Diseases: Viewing the 2013 Landscape', Available at:
http://c-path.org/wp-content/uploads/2013/11/lntro_11-12-2013-DStephenson.pdf.
Stephenson, D, Bain, L, Brumfield, M, Carrillo, M, Comery, T, Compton, C, Corrigan, B, Gordon, M, Jack,
C, Katz, R, Logovinsky, V, Satlin, A, Marek, K, Nicholas, T, Polhamus, D, Angersbach, B, Raghavan, N,
Romano, G, Romero, K, Shaw, L, Woodcock, J, Vradenburg, G & Isaac, M 2013, 'Coalition Against
Major Diseases: Precompetitive Collaborations and Regulatory Paths to Accelerating Drug
Development for Neurodegenerative Diseases'. Therapeutic Innovation & Regulatory Science,
47(6), pp.632-638.
Teutsch, SM, Bradley, L a, Palomaki, GE, Haddow, JE, Piper, M, Calonge, N, Dotson, WD, Douglas, MP &
Berg, AO 2009, 'The Evaluation of Genomic Applications in Practice and Prevention (EGAPP)
Initiative: methods of the EGAPP Working Group.'. Genetics in Medicine, 11(1), pp.3-14.
Trusheim, MR, Austin, MJF, Rausch, C & Berndt, ER 2013, 'Uncertain prognosis for high-quality
diagnostics: clinical challenges, economic barriers and needed reforms.'. Pharmacogenomics,
14(3), pp.325-34.
Trusheim, MR & Berndt, ER 2012, 'Economic challenges and possible policy actions to advance stratified
medicine'. Personalized Medicine, 9(4), pp.413-427.
Vaidya, VS & Bonventre, J V 2010, 'Biomarkers in Medicine, Drug Discovery, and Environmental Health',
Wiley.
Tingyu Liu 164
Vermillion 2013, 'Vermillion Reports Fourth Quarter and Full Year 2013 Results'. Vermillion. Available at:
http://phx.corporate-ir.net/phoenix.zhtml?c=121814&p=irol-newsArticle&lD=1906975&highlight=
[Accessed April 14, 2014].
Wagner, J 2008, 'Strategic approach to fit-for-purpose biomarkers in drug development.'. Annual Review
of Pharmacology and Toxicology, 48, pp.631-51.
Wagner, J, Prince, M, Wright, E, Ennis, M, Kochan, J, Nunez, D, Schneider, B, Wang, M, Chen, Y, Ghosh, S,
Musser, B & Vassileva, M 2010, 'The Biomarkers Consortium: practice and pitfalls of open-source
precompetitive collaboration'. Clinical Pharmacology & Therapeutics, 87(5), pp.539-542.
Wagner, J, Williams, S & Webster, C 2007, 'Biomarkers and surrogate end points for fit-for-purpose
development and regulatory evaluation of new drugs.'. Clinical Pharmacology & Therapeutics,
81(1), pp.104-7.
Wagner, J, Wright, E, Ennis, M, Prince, M, Kochan, J, Nunez, D, Schneider, B, Wang, M, Chen, Y, Ghosh, S,
Musser, B & Vassileva, M 2009, 'Utility of adiponectin as a biomarker predictive of glycemic
efficacy is demonstrated by collaborative pooling of data from clinical trials conducted by multiple
sponsors.'. Clinical Pharmacology & Therapeutics, 86(6), pp.619-25.
Wagner, P & Srivastava, S 2012, 'New paradigms in translational science research in cancer biomarkers'.
Translational Research, 159(4), pp.343-353.
Walker, EG, Brumfield, M, Compton, C & Woosley, R 2013, 'Evolving Global Regulatory Science Through
the Voluntary Submission of Data: A 2013 Assessment'. Therapeutic Innovation & Regulatory
Science.
Weiss, RL 2012, 'The long and winding regulatory road for laboratory-developed tests.'. American
Journal of Clinical Pathology, 138(1), pp.20-6.
Williams, PM, Lively, TG, Jessup, JM & Conley, BA 2012, 'Bridging the gap: moving predictive and
prognostic assays from research to clinical use.'. Clinical Cancer Research, 18(6), pp.1531-9.
Wilson, C, Schultz, S & Waldman, S 2007, 'Cancer biomarkers: where medicine, business, and public
policy intersect'. Biotechnology Healthcare, 4(1), pp.33-42.
Woodcock, J 2010, 'Assessing the Clinical Utility of Diagnostics Used in Drug Therapy'. Clinical
Pharmacology & Therapeutics, 88(6), pp.765-774.
Woodcock, J, Buckman, S, Goodsaid, F & Walton, M 2011, 'Qualifying biomarkers for use in drug
development: a US Food and Drug Administration overview'. Expert Opinion on Medical
Diagnostics, 5(5), pp.369-374.
Woodcock, J & Woosley, R 2008, 'The FDA critical path initiative and its influence on new drug
development.'. Annual Review of Medicine, 59, pp.1-12.
Tingyu Liu 165
Yu, H 2014, 'FDA weighs in on Drug Development Tools'. BioClinica. Available at:
http://www.bioclinica.com/blog/fda-weighs-drug-development-tools [Accessed April 25, 2014].
Zineh, I & Lesko, L 2009, 'Pharmacogenetics in medicine: barriers, critical factors and a framework for
dialogue'. Personalized Medicine, 6(4), pp.359-361.